public void Apply()
{
var pattern = Pattern <AnnotatedStringData, int> .New()
.Group("leftEnv", leftEnv => leftEnv
.Annotation(FeatureStruct.New(PhoneticFeatSys)
.Symbol(Seg)
.Symbol("cons+")
.Feature("voice").EqualToVariable("a").Value))
.Group("target", target => target
.Annotation(FeatureStruct.New(PhoneticFeatSys)
.Symbol(Seg)
.Symbol("cons-")
.Symbol("low+").Value))
.Group("rightEnv", rightEnv => rightEnv
.Annotation(FeatureStruct.New(PhoneticFeatSys)
.Symbol(Seg)
.Symbol("cons+")
.Feature("voice").Not.EqualToVariable("a").Value)).Value;
var ruleSpec = new DefaultPatternRuleSpec <AnnotatedStringData, int>(pattern, (r, match) =>
{
GroupCapture <int> target = match.GroupCaptures["target"];
foreach (Annotation <int> ann in match.Input.Annotations.GetNodes(target.Span))
{
ann.FeatureStruct.PriorityUnion(FeatureStruct.New(PhoneticFeatSys).Symbol("low-").Value);
}
return(match.Input);
});
var rule = new PatternRule <AnnotatedStringData, int>(SpanFactory, ruleSpec);
AnnotatedStringData inputWord = CreateStringData("fazk");
Assert.IsTrue(rule.Apply(inputWord).Any());
}
protected AnnotatedStringData CreateStringData(string str)
{
var stringData = new AnnotatedStringData(SpanFactory, str);
for (int i = 0; i < str.Length; i++)
{
FeatureStruct fs = Characters[str[i]];
stringData.Annotations.Add(i, i + 1, fs.DeepClone());
}
return(stringData);
}
public void Batch()
{
var pattern = Pattern <AnnotatedStringData, int> .New()
.Group("leftEnv", leftEnv => leftEnv
.Annotation(FeatureStruct.New(PhoneticFeatSys)
.Symbol(Seg)
.Symbol("cons+")
.Feature("voice").EqualToVariable("a").Value))
.Group("target", target => target
.Annotation(FeatureStruct.New(PhoneticFeatSys)
.Symbol(Seg)
.Symbol("cons-")
.Symbol("low+").Value))
.Group("rightEnv", rightEnv => rightEnv
.Annotation(FeatureStruct.New(PhoneticFeatSys)
.Symbol(Seg)
.Symbol("cons+")
.Feature("voice").Not.EqualToVariable("a").Value)).Value;
var ruleSpec1 = new DefaultPatternRuleSpec <AnnotatedStringData, int>(pattern, (r, match) =>
{
GroupCapture <int> target = match.GroupCaptures["target"];
foreach (Annotation <int> ann in match.Input.Annotations.GetNodes(target.Span))
{
ann.FeatureStruct.PriorityUnion(FeatureStruct.New(PhoneticFeatSys)
.Symbol("low-")
.Symbol("mid-").Value);
}
return(match.Input);
},
input => input.Annotations.Single(ann => ((FeatureSymbol)ann.FeatureStruct.GetValue(Type)) == Word)
.FeatureStruct.IsUnifiable(FeatureStruct.New(WordFeatSys).Symbol("verb").Value));
var ruleSpec2 = new DefaultPatternRuleSpec <AnnotatedStringData, int>(pattern, (r, match) =>
{
GroupCapture <int> target = match.GroupCaptures["target"];
foreach (Annotation <int> ann in match.Input.Annotations.GetNodes(target.Span))
{
ann.FeatureStruct.PriorityUnion(FeatureStruct.New(PhoneticFeatSys)
.Symbol("low-")
.Symbol("mid+").Value);
}
return(match.Input);
});
var batchSpec = new BatchPatternRuleSpec <AnnotatedStringData, int>(new[] { ruleSpec1, ruleSpec2 });
var rule = new PatternRule <AnnotatedStringData, int>(SpanFactory, batchSpec);
AnnotatedStringData inputWord = CreateStringData("fazk");
inputWord.Annotations.Add(inputWord.Span, FeatureStruct.New(WordFeatSys).Symbol(Word).Symbol("noun").Value);
Assert.IsTrue(rule.Apply(inputWord).Any());
}
public void NondeterministicPattern()
{
var any = FeatureStruct.New().Value;
var pattern = Pattern<AnnotatedStringData, int>.New()
.Group("first", first => first.Annotation(any).OneOrMore)
.Group("second", second => second.Annotation(any).OneOrMore).Value;
var matcher = new Matcher<AnnotatedStringData, int>(SpanFactory, pattern,
new MatcherSettings<int>
{
AnchoredToStart = true,
AnchoredToEnd = true,
AllSubmatches = true
});
var word = new AnnotatedStringData(SpanFactory, "test");
word.Annotations.Add(0, 1, FeatureStruct.New(PhoneticFeatSys).Feature("strRep").EqualTo("t").Value);
word.Annotations.Add(1, 2, FeatureStruct.New(PhoneticFeatSys).Feature("strRep").EqualTo("e").Value);
word.Annotations.Add(2, 3, FeatureStruct.New(PhoneticFeatSys).Feature("strRep").EqualTo("s").Value);
word.Annotations.Add(3, 4, FeatureStruct.New(PhoneticFeatSys).Feature("strRep").EqualTo("t").Value);
Match<AnnotatedStringData, int>[] matches = matcher.AllMatches(word).ToArray();
Assert.AreEqual(3, matches.Length);
Assert.AreEqual(SpanFactory.Create(0, 4), matches[0].Span);
Assert.AreEqual(SpanFactory.Create(0, 3), matches[0].GroupCaptures["first"].Span);
Assert.AreEqual(SpanFactory.Create(3, 4), matches[0].GroupCaptures["second"].Span);
Assert.AreEqual(SpanFactory.Create(0, 4), matches[1].Span);
Assert.AreEqual(SpanFactory.Create(0, 2), matches[1].GroupCaptures["first"].Span);
Assert.AreEqual(SpanFactory.Create(2, 4), matches[1].GroupCaptures["second"].Span);
Assert.AreEqual(SpanFactory.Create(0, 4), matches[2].Span);
Assert.AreEqual(SpanFactory.Create(0, 1), matches[2].GroupCaptures["first"].Span);
Assert.AreEqual(SpanFactory.Create(1, 4), matches[2].GroupCaptures["second"].Span);
}
protected AnnotatedStringData CreateStringData(string str)
{
var stringData = new AnnotatedStringData(SpanFactory, str);
for (int i = 0; i < str.Length; i++)
{
FeatureStruct fs = Characters[str[i]];
stringData.Annotations.Add(i, i + 1, fs.DeepClone());
}
return stringData;
}
public void NondeterministicPattern()
{
var any = FeatureStruct.New().Value;
var pattern = Pattern<AnnotatedStringData, int>.New()
.Group("first", first => first.Annotation(any).OneOrMore)
.Group("second", second => second.Annotation(any).OneOrMore).Value;
var matcher = new Matcher<AnnotatedStringData, int>(SpanFactory, pattern,
new MatcherSettings<int>
{
AnchoredToStart = true,
AnchoredToEnd = true,
AllSubmatches = true
});
var word = new AnnotatedStringData(SpanFactory, "test");
word.Annotations.Add(0, 1, FeatureStruct.New(PhoneticFeatSys).Feature("strRep").EqualTo("t").Value);
word.Annotations.Add(1, 2, FeatureStruct.New(PhoneticFeatSys).Feature("strRep").EqualTo("e").Value);
word.Annotations.Add(2, 3, FeatureStruct.New(PhoneticFeatSys).Feature("strRep").EqualTo("s").Value);
word.Annotations.Add(3, 4, FeatureStruct.New(PhoneticFeatSys).Feature("strRep").EqualTo("t").Value);
Match<AnnotatedStringData, int>[] matches = matcher.AllMatches(word).ToArray();
Assert.AreEqual(3, matches.Length);
Assert.AreEqual(SpanFactory.Create(0, 4), matches[0].Span);
Assert.AreEqual(SpanFactory.Create(0, 3), matches[0].GroupCaptures["first"].Span);
Assert.AreEqual(SpanFactory.Create(3, 4), matches[0].GroupCaptures["second"].Span);
Assert.AreEqual(SpanFactory.Create(0, 4), matches[1].Span);
Assert.AreEqual(SpanFactory.Create(0, 2), matches[1].GroupCaptures["first"].Span);
Assert.AreEqual(SpanFactory.Create(2, 4), matches[1].GroupCaptures["second"].Span);
Assert.AreEqual(SpanFactory.Create(0, 4), matches[2].Span);
Assert.AreEqual(SpanFactory.Create(0, 1), matches[2].GroupCaptures["first"].Span);
Assert.AreEqual(SpanFactory.Create(1, 4), matches[2].GroupCaptures["second"].Span);
pattern = Pattern<AnnotatedStringData, int>.New()
.Group("first", g1 => g1.Annotation(FeatureStruct.New(PhoneticFeatSys).Symbol(Seg).Symbol("syl+").Value))
.Group("second", g2 => g2.Group("third", g3 => g3.Annotation(FeatureStruct.New().Symbol(Seg).Value).Optional).ZeroOrMore).Value;
matcher = new Matcher<AnnotatedStringData, int>(SpanFactory, pattern,
new MatcherSettings<int>
{
AnchoredToStart = true,
AnchoredToEnd = true,
AllSubmatches = true
});
word = CreateStringData("etested");
matches = matcher.AllMatches(word).ToArray();
Assert.That(matches.Length, Is.EqualTo(1));
Assert.That(matches[0].Success, Is.True);
Assert.That(matches[0].Span, Is.EqualTo(SpanFactory.Create(0, 7)));
Assert.That(matches[0].GroupCaptures["first"].Span, Is.EqualTo(SpanFactory.Create(0, 1)));
Assert.That(matches[0].GroupCaptures["second"].Span, Is.EqualTo(SpanFactory.Create(1, 7)));
Assert.That(matches[0].GroupCaptures["third"].Span, Is.EqualTo(SpanFactory.Create(6, 7)));
word = CreateStringData("e");
matches = matcher.AllMatches(word).ToArray();
Assert.That(matches.Length, Is.EqualTo(1));
Assert.That(matches[0].Success, Is.True);
Assert.That(matches[0].Span, Is.EqualTo(SpanFactory.Create(0, 1)));
Assert.That(matches[0].GroupCaptures["first"].Span, Is.EqualTo(SpanFactory.Create(0, 1)));
Assert.That(matches[0].GroupCaptures["second"].Success, Is.False);
Assert.That(matches[0].GroupCaptures["third"].Success, Is.False);
pattern = Pattern<AnnotatedStringData, int>.New()
.Annotation(FeatureStruct.New(PhoneticFeatSys).Symbol(Seg).Symbol("syl+").Value)
.Group("first", g => g.Annotation(FeatureStruct.New().Symbol(Seg).Value).ZeroOrMore).Value;
matcher = new Matcher<AnnotatedStringData, int>(SpanFactory, pattern,
new MatcherSettings<int>
{
AnchoredToStart = true,
AnchoredToEnd = true,
AllSubmatches = true
});
word = CreateStringData("etested");
matches = matcher.AllMatches(word).ToArray();
Assert.That(matches.Length, Is.EqualTo(1));
Assert.That(matches[0].Success, Is.True);
Assert.That(matches[0].Span, Is.EqualTo(SpanFactory.Create(0, 7)));
Assert.That(matches[0].GroupCaptures["first"].Span, Is.EqualTo(SpanFactory.Create(1, 7)));
word = CreateStringData("e");
matches = matcher.AllMatches(word).ToArray();
Assert.That(matches.Length, Is.EqualTo(1));
Assert.That(matches[0].Success, Is.True);
Assert.That(matches[0].Span, Is.EqualTo(SpanFactory.Create(0, 1)));
Assert.That(matches[0].GroupCaptures["first"].Success, Is.False);
}
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" }));
}
