private int Diff(FeatureStruct fs1, FeatureStruct fs2, SymbolicFeature feature)
{
SymbolicFeatureValue pValue;
if (!fs1.TryGetValue(feature, out pValue))
{
return(0);
}
SymbolicFeatureValue qValue;
if (!fs2.TryGetValue(feature, out qValue))
{
return(0);
}
int sum = 0;
int count = 0;
foreach (FeatureSymbol pSymbol in pValue.Values)
{
foreach (FeatureSymbol qSymbol in qValue.Values)
{
sum += Math.Abs(_valueMetrics[pSymbol] - _valueMetrics[qSymbol]);
count++;
}
}
return(sum / count);
}
static CogFeatureSystem()
{
AnchorType = new FeatureSymbol(Guid.NewGuid().ToString()) {Description = "anchor"};
VowelType = new FeatureSymbol(Guid.NewGuid().ToString()) {Description = "vowel"};
ConsonantType = new FeatureSymbol(Guid.NewGuid().ToString()) {Description = "consonant"};
ToneLetterType = new FeatureSymbol(Guid.NewGuid().ToString()) {Description = "toneLetter"};
BoundaryType = new FeatureSymbol(Guid.NewGuid().ToString()) {Description = "boundary"};
SyllableType = new FeatureSymbol(Guid.NewGuid().ToString()) {Description = "syllable"};
StemType = new FeatureSymbol(Guid.NewGuid().ToString()) {Description = "stem"};
PrefixType = new FeatureSymbol(Guid.NewGuid().ToString()) {Description = "prefix"};
SuffixType = new FeatureSymbol(Guid.NewGuid().ToString()) {Description = "suffix"};
Type = new SymbolicFeature(Guid.NewGuid().ToString(), AnchorType, VowelType, ConsonantType, ToneLetterType, BoundaryType, SyllableType, StemType, PrefixType, SuffixType) {Description = "Type"};
Complex = new FeatureSymbol(Guid.NewGuid().ToString()) {Description = "complex"};
Simple = new FeatureSymbol(Guid.NewGuid().ToString()) {Description = "simple"};
SegmentType = new SymbolicFeature(Guid.NewGuid().ToString(), Complex, Simple) {Description = "SegmentType"};
OriginalStrRep = new StringFeature(Guid.NewGuid().ToString()) {Description = "OriginalStrRep"};
StrRep = new StringFeature(Guid.NewGuid().ToString()) {Description = "StrRep"};
First = new ComplexFeature(Guid.NewGuid().ToString()) {Description = "First"};
Onset = new FeatureSymbol(Guid.NewGuid().ToString()) {Description = "onset"};
Nucleus = new FeatureSymbol(Guid.NewGuid().ToString()) {Description = "nucleus"};
Coda = new FeatureSymbol(Guid.NewGuid().ToString()) {Description = "coda"};
SyllablePosition = new SymbolicFeature(Guid.NewGuid().ToString(), Onset, Nucleus, Coda) {Description = "SyllablePosition"};
Instance = new CogFeatureSystem();
}
private int Diff(FeatureStruct fs1, FeatureStruct fs2, SymbolicFeature feature)
{
SymbolicFeatureValue pValue;
if (!fs1.TryGetValue(feature, out pValue))
{
pValue = null;
}
SymbolicFeatureValue qValue;
if (!fs2.TryGetValue(feature, out qValue))
{
qValue = null;
}
if (pValue == null && qValue == null)
{
return(0);
}
FeatureSymbol[] values1 = pValue == null?feature.PossibleSymbols.ToArray() : pValue.Values.ToArray();
FeatureSymbol[] values2 = qValue == null?feature.PossibleSymbols.ToArray() : qValue.Values.ToArray();
if (values2.Length > values1.Length)
{
FeatureSymbol[] temp = values1;
values1 = values2;
values2 = temp;
}
return((int)Math.Round(values1.Average(s1 => values2.Min(s2 => Math.Abs(_valueMetrics[s1] - _valueMetrics[s2])))));
}
public FeatureViewModel(SymbolicFeature feature, FeatureSymbol symbol)
{
_feature = feature;
_values = new ReadOnlyCollection<FeatureValueViewModel>(feature.PossibleSymbols.Select(s => new FeatureValueViewModel(s)).ToArray());
if (symbol != null)
_selectedValue = _values.Single(fv => fv.DomainSymbol == symbol);
else if (_values.Count > 0)
_selectedValue = _values[0];
}
public SymbolicFeature AddPartsOfSpeech(IEnumerable <FeatureSymbol> pos)
{
var posFeature = new SymbolicFeature(PartOfSpeechID, pos)
{
Description = "POS"
};
base.Add(posFeature);
return(posFeature);
}
public void Migrate(SegmentPool segmentPool, CogProject project)
{
if (project.FeatureSystem.ContainsFeature("airstream"))
return;
var pulmonic = new FeatureSymbol("pulmonic");
var ejective = new FeatureSymbol("ejective");
var implosive = new FeatureSymbol("implosive");
var click = new FeatureSymbol("click");
var airstream = new SymbolicFeature("airstream",
pulmonic,
ejective,
implosive,
click) {Description = "Airstream"};
project.FeatureSystem.Add(airstream);
AddValue(project.Segmenter.Modifiers, "ʼ", ejective);
AddValue(project.Segmenter.Consonants, "ɓ", implosive);
AddValue(project.Segmenter.Consonants, "ɗ", implosive);
AddValue(project.Segmenter.Consonants, "ʄ", implosive);
AddValue(project.Segmenter.Consonants, "ɠ", implosive);
AddValue(project.Segmenter.Consonants, "ʛ", implosive);
FeatureSymbol affricate;
if (project.FeatureSystem.TryGetSymbol("affricate", out affricate))
{
project.Segmenter.Consonants.AddSymbolBasedOn("ʘ", "p", affricate, click);
project.Segmenter.Consonants.AddSymbolBasedOn("ǀ", "θ", affricate, click);
project.Segmenter.Consonants.AddSymbolBasedOn("ǁ", "ɬ", affricate, click);
}
project.Segmenter.Consonants.AddSymbolBasedOn("ǃ", "t", click);
project.Segmenter.Consonants.AddSymbolBasedOn("ǂ", "c", click);
foreach (Symbol symbol in project.Segmenter.Vowels.ToArray())
AddValue(project.Segmenter.Vowels, symbol, pulmonic);
foreach (Symbol symbol in project.Segmenter.Consonants.Where(s => !s.FeatureStruct.ContainsFeature("airstream")).ToArray())
AddValue(project.Segmenter.Consonants, symbol, pulmonic);
foreach (KeyValuePair<string, IWordAligner> aligner in project.WordAligners.Where(kvp => kvp.Value is Aline).ToArray())
{
var aline = (Aline) aligner.Value;
Dictionary<SymbolicFeature, int> featWeights = aline.FeatureWeights.ToDictionary(kvp => kvp.Key, kvp => kvp.Value);
featWeights[airstream] = 5;
Dictionary<FeatureSymbol, int> valueMetrics = aline.ValueMetrics.ToDictionary(kvp => kvp.Key, kvp => kvp.Value);
valueMetrics[pulmonic] = 100;
valueMetrics[ejective] = 66;
valueMetrics[implosive] = 33;
valueMetrics[click] = 0;
project.WordAligners[aligner.Key] = new Aline(segmentPool, aline.RelevantVowelFeatures, aline.RelevantConsonantFeatures.Concat(airstream),
featWeights, valueMetrics, aline.Settings);
}
}
public FeatureViewModel(SymbolicFeature feature, FeatureSymbol symbol)
{
_feature = feature;
_values = new ReadOnlyCollection <FeatureValueViewModel>(feature.PossibleSymbols.Select(s => new FeatureValueViewModel(s)).ToArray());
if (symbol != null)
{
_selectedValue = _values.Single(fv => fv.DomainSymbol == symbol);
}
else if (_values.Count > 0)
{
_selectedValue = _values[0];
}
}
public RelevantFeatureViewModel(SymbolicFeature feature, int weight, bool vowel, bool consonant, IReadOnlyDictionary<FeatureSymbol, int> valueMetrics)
{
_feature = feature;
_weight = weight;
_vowel = vowel;
_consonant = consonant;
var values = new List<RelevantValueViewModel>();
foreach (FeatureSymbol symbol in feature.PossibleSymbols)
{
var vm = new RelevantValueViewModel(symbol, valueMetrics[symbol]);
vm.PropertyChanged += ChildPropertyChanged;
values.Add(vm);
}
_values = new ReadOnlyList<RelevantValueViewModel>(values);
}
public RelevantFeatureViewModel(SymbolicFeature feature, int weight, bool vowel, bool consonant, IReadOnlyDictionary <FeatureSymbol, int> valueMetrics)
{
_feature = feature;
_weight = weight;
_vowel = vowel;
_consonant = consonant;
var values = new List <RelevantValueViewModel>();
foreach (FeatureSymbol symbol in feature.PossibleSymbols)
{
var vm = new RelevantValueViewModel(symbol, valueMetrics[symbol]);
vm.PropertyChanged += ChildPropertyChanged;
values.Add(vm);
}
_values = new ReadOnlyList <RelevantValueViewModel>(values);
}
static HCFeatureSystem()
{
Anchor = new FeatureSymbol(Guid.NewGuid().ToString()) { Description = "anchor" };
Segment = new FeatureSymbol(Guid.NewGuid().ToString()) { Description = "segment" };
Boundary = new FeatureSymbol(Guid.NewGuid().ToString()) { Description = "boundary" };
Morph = new FeatureSymbol(Guid.NewGuid().ToString()) { Description = "morph" };
Type = new SymbolicFeature(Guid.NewGuid().ToString(), Anchor, Segment, Boundary, Morph) { Description = "Type" };
Dirty = new FeatureSymbol(Guid.NewGuid().ToString()) { Description = "Dirty" };
Clean = new FeatureSymbol(Guid.NewGuid().ToString()) { Description = "Clean" };
Modified = new SymbolicFeature(Guid.NewGuid().ToString(), Dirty, Clean)
{
Description = "Modified",
DefaultValue = new SymbolicFeatureValue(Clean)
};
Deleted = new FeatureSymbol(Guid.NewGuid().ToString()) {Description = "Deleted"};
NotDeleted = new FeatureSymbol(Guid.NewGuid().ToString()) {Description = "NotDeleted"};
Deletion = new SymbolicFeature(Guid.NewGuid().ToString(), Deleted, NotDeleted)
{
Description = "Deletion",
DefaultValue = new SymbolicFeatureValue(NotDeleted)
};
LeftSide = new FeatureSymbol(Guid.NewGuid().ToString()) { Description = "LeftSide" };
RightSide = new FeatureSymbol(Guid.NewGuid().ToString()) { Description = "RightSide" };
AnchorType = new SymbolicFeature(Guid.NewGuid().ToString(), LeftSide, RightSide) { Description = "AnchorType" };
StrRep = new StringFeature(Guid.NewGuid().ToString()) {Description = "StrRep"};
Allomorph = new StringFeature(Guid.NewGuid().ToString()) {Description = "Allomorph"};
Instance = new HCFeatureSystem();
LeftSideAnchor = FeatureStruct.New().Symbol(Anchor).Symbol(LeftSide).Value;
RightSideAnchor = FeatureStruct.New().Symbol(Anchor).Symbol(RightSide).Value;
}
public FeatureViewModel(SymbolicFeature feature)
: this(feature, null)
{
}
public SymbolicFeature AddPartsOfSpeech(IEnumerable<FeatureSymbol> pos)
{
var posFeature = new SymbolicFeature(PartOfSpeechID, pos) {Description = "POS"};
base.Add(posFeature);
return posFeature;
}
public FeatureViewModel(SymbolicFeature feature)
: this(feature, null)
{
}
static HCFeatureSystem()
{
Anchor = new FeatureSymbol(Guid.NewGuid().ToString())
{
Description = "anchor"
};
Segment = new FeatureSymbol(Guid.NewGuid().ToString())
{
Description = "segment"
};
Boundary = new FeatureSymbol(Guid.NewGuid().ToString())
{
Description = "boundary"
};
Morph = new FeatureSymbol(Guid.NewGuid().ToString())
{
Description = "morph"
};
Type = new SymbolicFeature(Guid.NewGuid().ToString(), Anchor, Segment, Boundary, Morph)
{
Description = "Type"
};
Dirty = new FeatureSymbol(Guid.NewGuid().ToString())
{
Description = "Dirty"
};
Clean = new FeatureSymbol(Guid.NewGuid().ToString())
{
Description = "Clean"
};
Modified = new SymbolicFeature(Guid.NewGuid().ToString(), Dirty, Clean)
{
Description = "Modified",
DefaultValue = new SymbolicFeatureValue(Clean)
};
Deleted = new FeatureSymbol(Guid.NewGuid().ToString())
{
Description = "Deleted"
};
NotDeleted = new FeatureSymbol(Guid.NewGuid().ToString())
{
Description = "NotDeleted"
};
Deletion = new SymbolicFeature(Guid.NewGuid().ToString(), Deleted, NotDeleted)
{
Description = "Deletion",
DefaultValue = new SymbolicFeatureValue(NotDeleted)
};
LeftSide = new FeatureSymbol(Guid.NewGuid().ToString())
{
Description = "LeftSide"
};
RightSide = new FeatureSymbol(Guid.NewGuid().ToString())
{
Description = "RightSide"
};
AnchorType = new SymbolicFeature(Guid.NewGuid().ToString(), LeftSide, RightSide)
{
Description = "AnchorType"
};
StrRep = new StringFeature(Guid.NewGuid().ToString())
{
Description = "StrRep"
};
Allomorph = new StringFeature(Guid.NewGuid().ToString())
{
Description = "Allomorph"
};
Instance = new HCFeatureSystem();
LeftSideAnchor = FeatureStruct.New().Symbol(Anchor).Symbol(LeftSide).Value;
RightSideAnchor = FeatureStruct.New().Symbol(Anchor).Symbol(RightSide).Value;
}
public virtual void FixtureSetUp()
{
SpanFactory = new IntegerSpanFactory();
PhoneticFeatSys = new FeatureSystem
{
new SymbolicFeature("son", new FeatureSymbol("son+", "+"), new FeatureSymbol("son-", "-"), new FeatureSymbol("son?", "?"))
{
DefaultSymbolID = "son?"
},
new SymbolicFeature("syl", new FeatureSymbol("syl+", "+"), new FeatureSymbol("syl-", "-"), new FeatureSymbol("syl?", "?"))
{
DefaultSymbolID = "syl?"
},
new SymbolicFeature("cons", new FeatureSymbol("cons+", "+"), new FeatureSymbol("cons-", "-"), new FeatureSymbol("cons?", "?"))
{
DefaultSymbolID = "cons?"
},
new SymbolicFeature("high", new FeatureSymbol("high+", "+"), new FeatureSymbol("high-", "-"), new FeatureSymbol("high?", "?"))
{
DefaultSymbolID = "high?"
},
new SymbolicFeature("back", new FeatureSymbol("back+", "+"), new FeatureSymbol("back-", "-"), new FeatureSymbol("back?", "?"))
{
DefaultSymbolID = "back?"
},
new SymbolicFeature("front", new FeatureSymbol("front+", "+"), new FeatureSymbol("front-", "-"), new FeatureSymbol("front?", "?"))
{
DefaultSymbolID = "front?"
},
new SymbolicFeature("low", new FeatureSymbol("low+", "+"), new FeatureSymbol("low-", "-"), new FeatureSymbol("low?", "?"))
{
DefaultSymbolID = "low?"
},
new SymbolicFeature("rnd", new FeatureSymbol("rnd+", "+"), new FeatureSymbol("rnd-", "-"), new FeatureSymbol("rnd?", "?"))
{
DefaultSymbolID = "rnd?"
},
new SymbolicFeature("ant", new FeatureSymbol("ant+", "+"), new FeatureSymbol("ant-", "-"), new FeatureSymbol("ant?", "?"))
{
DefaultSymbolID = "ant?"
},
new SymbolicFeature("cor", new FeatureSymbol("cor+", "+"), new FeatureSymbol("cor-", "-"), new FeatureSymbol("cor?", "?"))
{
DefaultSymbolID = "cor?"
},
new SymbolicFeature("voice", new FeatureSymbol("voice+", "+"), new FeatureSymbol("voice-", "-"), new FeatureSymbol("voice?", "?"))
{
DefaultSymbolID = "voice?"
},
new SymbolicFeature("cont", new FeatureSymbol("cont+", "+"), new FeatureSymbol("cont-", "-"), new FeatureSymbol("cont?", "?"))
{
DefaultSymbolID = "cont?"
},
new SymbolicFeature("nas", new FeatureSymbol("nas+", "+"), new FeatureSymbol("nas-", "-"), new FeatureSymbol("nas?", "?"))
{
DefaultSymbolID = "nas?"
},
new SymbolicFeature("str", new FeatureSymbol("str+", "+"), new FeatureSymbol("str-", "-"), new FeatureSymbol("str?", "?"))
{
DefaultSymbolID = "str?"
},
new StringFeature("strRep")
};
WordFeatSys = new FeatureSystem
{
new SymbolicFeature("POS", new FeatureSymbol("noun"), new FeatureSymbol("verb"), new FeatureSymbol("adj"), new FeatureSymbol("adv"), new FeatureSymbol("det"))
};
Word = new FeatureSymbol("Word");
NP = new FeatureSymbol("NP");
VP = new FeatureSymbol("VP");
Seg = new FeatureSymbol("Seg");
Bdry = new FeatureSymbol("Bdry");
Allo = new FeatureSymbol("Allo");
Type = new SymbolicFeature("Type", Word, NP, VP, Seg, Bdry, Allo);
TypeFeatSys = new FeatureSystem {
Type
};
Characters = new Dictionary <char, FeatureStruct>
{
{ 'b', FeatureStruct.New(PhoneticFeatSys)
.Symbol(Seg)
.Symbol("son-")
.Symbol("syl-")
.Symbol("cons+")
.Symbol("high-")
.Symbol("ant+")
.Symbol("cor-")
.Symbol("voice+")
.Symbol("cont-")
.Symbol("nas-")
.Symbol("str-").Value },
{ 'd', FeatureStruct.New(PhoneticFeatSys)
.Symbol(Seg)
.Symbol("son-")
.Symbol("syl-")
.Symbol("cons+")
.Symbol("high-")
.Symbol("ant+")
.Symbol("cor+")
.Symbol("voice+")
.Symbol("cont-")
.Symbol("nas-")
.Symbol("str-").Value },
{ 'g', FeatureStruct.New(PhoneticFeatSys)
.Symbol(Seg)
.Symbol("son-")
.Symbol("syl-")
.Symbol("cons+")
.Symbol("high+")
.Symbol("ant-")
.Symbol("cor-")
.Symbol("voice+")
.Symbol("cont-")
.Symbol("nas-")
.Symbol("str-").Value },
{ 'p', FeatureStruct.New(PhoneticFeatSys)
.Symbol(Seg)
.Symbol("son-")
.Symbol("syl-")
.Symbol("cons+")
.Symbol("high-")
.Symbol("ant+")
.Symbol("cor-")
.Symbol("voice-")
.Symbol("cont-")
.Symbol("nas-")
.Symbol("str-").Value },
{ 't', FeatureStruct.New(PhoneticFeatSys)
.Symbol(Seg)
.Symbol("son-")
.Symbol("syl-")
.Symbol("cons+")
.Symbol("high-")
.Symbol("ant+")
.Symbol("cor+")
.Symbol("voice-")
.Symbol("cont-")
.Symbol("nas-")
.Symbol("str-").Value },
{ 'q', FeatureStruct.New(PhoneticFeatSys)
.Symbol(Seg)
.Symbol("son-")
.Symbol("syl-")
.Symbol("cons+")
.Symbol("high+")
.Symbol("ant-")
.Symbol("cor-")
.Symbol("voice-")
.Symbol("cont-")
.Symbol("nas-")
.Symbol("str-").Value },
{ 'c', FeatureStruct.New(PhoneticFeatSys)
.Symbol(Seg)
.Symbol("son-")
.Symbol("syl-")
.Symbol("cons+")
.Symbol("high+")
.Symbol("ant-")
.Symbol("cor-")
.Symbol("voice-")
.Symbol("cont-")
.Symbol("nas-")
.Symbol("str-").Value },
{ 'k', FeatureStruct.New(PhoneticFeatSys)
.Symbol(Seg)
.Symbol("son-")
.Symbol("syl-")
.Symbol("cons+")
.Symbol("high+")
.Symbol("ant-")
.Symbol("cor-")
.Symbol("voice-")
.Symbol("cont-")
.Symbol("nas-")
.Symbol("str-").Value },
{ 'x', FeatureStruct.New(PhoneticFeatSys)
.Symbol(Seg)
.Symbol("son-")
.Symbol("syl-")
.Symbol("cons+")
.Symbol("high+")
.Symbol("ant-")
.Symbol("cor-")
.Symbol("voice-")
.Symbol("cont-")
.Symbol("nas-")
.Symbol("str-").Value },
{ 'j', FeatureStruct.New(PhoneticFeatSys)
.Symbol(Seg)
.Symbol("son-")
.Symbol("syl-")
.Symbol("cons+")
.Symbol("high+")
.Symbol("ant-")
.Symbol("cor+")
.Symbol("voice+")
.Symbol("cont-")
.Symbol("nas-")
.Symbol("str+").Value },
{ 's', FeatureStruct.New(PhoneticFeatSys)
.Symbol(Seg)
.Symbol("son-")
.Symbol("syl-")
.Symbol("cons+")
.Symbol("high-")
.Symbol("ant+")
.Symbol("cor+")
.Symbol("voice-")
.Symbol("cont+")
.Symbol("nas-")
.Symbol("str+").Value },
{ 'z', FeatureStruct.New(PhoneticFeatSys)
.Symbol(Seg)
.Symbol("son-")
.Symbol("syl-")
.Symbol("cons+")
.Symbol("high-")
.Symbol("ant+")
.Symbol("cor+")
.Symbol("voice+")
.Symbol("cont+")
.Symbol("nas-")
.Symbol("str+").Value },
{ 'f', FeatureStruct.New(PhoneticFeatSys)
.Symbol(Seg)
.Symbol("son-")
.Symbol("syl-")
.Symbol("cons+")
.Symbol("high-")
.Symbol("ant+")
.Symbol("cor-")
.Symbol("voice-")
.Symbol("cont+")
.Symbol("nas-")
.Symbol("str+").Value },
{ 'v', FeatureStruct.New(PhoneticFeatSys)
.Symbol(Seg)
.Symbol("son-")
.Symbol("syl-")
.Symbol("cons+")
.Symbol("high-")
.Symbol("ant+")
.Symbol("cor-")
.Symbol("voice+")
.Symbol("cont+")
.Symbol("nas-")
.Symbol("str+").Value },
{ 'w', FeatureStruct.New(PhoneticFeatSys)
.Symbol(Seg)
.Symbol("son+")
.Symbol("syl-")
.Symbol("cons-")
.Symbol("high+")
.Symbol("back+")
.Symbol("front-")
.Symbol("low-")
.Symbol("rnd+")
.Symbol("ant-")
.Symbol("cor-").Value },
{ 'y', FeatureStruct.New(PhoneticFeatSys)
.Symbol(Seg)
.Symbol("son+")
.Symbol("syl-")
.Symbol("cons-")
.Symbol("high+")
.Symbol("back-")
.Symbol("front+")
.Symbol("low-")
.Symbol("rnd-")
.Symbol("ant-")
.Symbol("cor-").Value },
{ 'h', FeatureStruct.New(PhoneticFeatSys)
.Symbol(Seg)
.Symbol("son+")
.Symbol("syl-")
.Symbol("cons-")
.Symbol("high-")
.Symbol("back-")
.Symbol("front-")
.Symbol("low+")
.Symbol("ant-")
.Symbol("cor-")
.Symbol("voice-")
.Symbol("cont+")
.Symbol("nas-")
.Symbol("str-").Value },
{ 'r', FeatureStruct.New(PhoneticFeatSys)
.Symbol(Seg)
.Symbol("son+")
.Symbol("syl-")
.Symbol("cons+")
.Symbol("high-")
.Symbol("back-")
.Symbol("front-")
.Symbol("low-")
.Symbol("ant-")
.Symbol("cor+")
.Symbol("voice+")
.Symbol("cont+")
.Symbol("nas-")
.Symbol("str-").Value },
{ 'l', FeatureStruct.New(PhoneticFeatSys)
.Symbol(Seg)
.Symbol("son+")
.Symbol("syl-")
.Symbol("cons+")
.Symbol("high-")
.Symbol("back-")
.Symbol("front-")
.Symbol("low-")
.Symbol("ant+")
.Symbol("cor+")
.Symbol("voice+")
.Symbol("cont+")
.Symbol("nas-")
.Symbol("str-").Value },
{ 'm', FeatureStruct.New(PhoneticFeatSys)
.Symbol(Seg)
.Symbol("son+")
.Symbol("syl-")
.Symbol("cons+")
.Symbol("high-")
.Symbol("low-")
.Symbol("ant+")
.Symbol("cor-")
.Symbol("voice+")
.Symbol("cont-")
.Symbol("nas+")
.Symbol("str-").Value },
{ 'n', FeatureStruct.New(PhoneticFeatSys)
.Symbol(Seg)
.Symbol("son+")
.Symbol("syl-")
.Symbol("cons+")
.Symbol("high-")
.Symbol("low-")
.Symbol("ant+")
.Symbol("cor+")
.Symbol("voice+")
.Symbol("cont-")
.Symbol("nas+")
.Symbol("str-").Value },
{ 'N', FeatureStruct.New(PhoneticFeatSys)
.Symbol(Seg)
.Symbol("son+")
.Symbol("syl-")
.Symbol("cons+")
.Symbol("high-")
.Symbol("low-")
.Symbol("ant+")
.Symbol("cor?")
.Symbol("voice+")
.Symbol("cont-")
.Symbol("nas+")
.Symbol("str-").Value },
{ 'a', FeatureStruct.New(PhoneticFeatSys)
.Symbol(Seg)
.Symbol("son+")
.Symbol("syl+")
.Symbol("cons-")
.Symbol("high-")
.Symbol("back-")
.Symbol("front+")
.Symbol("low+")
.Symbol("rnd-").Value },
{ 'e', FeatureStruct.New(PhoneticFeatSys)
.Symbol(Seg)
.Symbol("son+")
.Symbol("syl+")
.Symbol("cons-")
.Symbol("high-")
.Symbol("back-")
.Symbol("front+")
.Symbol("low-")
.Symbol("rnd-").Value },
{ 'i', FeatureStruct.New(PhoneticFeatSys)
.Symbol(Seg)
.Symbol("son+")
.Symbol("syl+")
.Symbol("cons-")
.Symbol("high+")
.Symbol("back-")
.Symbol("front+")
.Symbol("low-")
.Symbol("rnd-").Value },
{ 'o', FeatureStruct.New(PhoneticFeatSys)
.Symbol(Seg)
.Symbol("son+")
.Symbol("syl+")
.Symbol("cons-")
.Symbol("high-")
.Symbol("back+")
.Symbol("front-")
.Symbol("low+")
.Symbol("rnd-").Value },
{ 'u', FeatureStruct.New(PhoneticFeatSys)
.Symbol(Seg)
.Symbol("son+")
.Symbol("syl+")
.Symbol("cons-")
.Symbol("high+")
.Symbol("back+")
.Symbol("front-")
.Symbol("low-")
.Symbol("rnd+").Value },
{ '+', FeatureStruct.New(PhoneticFeatSys)
.Symbol(Bdry)
.Feature("strRep").EqualTo("+").Value },
{ ',', FeatureStruct.New(PhoneticFeatSys)
.Symbol(Bdry)
.Feature("strRep").EqualTo(",").Value },
{ ' ', FeatureStruct.New(PhoneticFeatSys)
.Symbol(Bdry)
.Feature("strRep").EqualTo(" ").Value },
{ '.', FeatureStruct.New(PhoneticFeatSys)
.Symbol(Bdry)
.Feature("strRep").EqualTo(".").Value }
};
}
public virtual void FixtureSetUp()
{
SpanFactory = new IntegerSpanFactory();
PhoneticFeatSys = new FeatureSystem
{
new SymbolicFeature("son", new FeatureSymbol("son+", "+"), new FeatureSymbol("son-", "-"), new FeatureSymbol("son?", "?")) {DefaultSymbolID = "son?"},
new SymbolicFeature("syl", new FeatureSymbol("syl+", "+"), new FeatureSymbol("syl-", "-"), new FeatureSymbol("syl?", "?")) {DefaultSymbolID = "syl?"},
new SymbolicFeature("cons", new FeatureSymbol("cons+", "+"), new FeatureSymbol("cons-", "-"), new FeatureSymbol("cons?", "?")) {DefaultSymbolID = "cons?"},
new SymbolicFeature("high", new FeatureSymbol("high+", "+"), new FeatureSymbol("high-", "-"), new FeatureSymbol("high?", "?")) {DefaultSymbolID = "high?"},
new SymbolicFeature("back", new FeatureSymbol("back+", "+"), new FeatureSymbol("back-", "-"), new FeatureSymbol("back?", "?")) {DefaultSymbolID = "back?"},
new SymbolicFeature("front", new FeatureSymbol("front+", "+"), new FeatureSymbol("front-", "-"), new FeatureSymbol("front?", "?")) {DefaultSymbolID = "front?"},
new SymbolicFeature("low", new FeatureSymbol("low+", "+"), new FeatureSymbol("low-", "-"), new FeatureSymbol("low?", "?")) {DefaultSymbolID = "low?"},
new SymbolicFeature("rnd", new FeatureSymbol("rnd+", "+"), new FeatureSymbol("rnd-", "-"), new FeatureSymbol("rnd?", "?")) {DefaultSymbolID = "rnd?"},
new SymbolicFeature("ant", new FeatureSymbol("ant+", "+"), new FeatureSymbol("ant-", "-"), new FeatureSymbol("ant?", "?")) {DefaultSymbolID = "ant?"},
new SymbolicFeature("cor", new FeatureSymbol("cor+", "+"), new FeatureSymbol("cor-", "-"), new FeatureSymbol("cor?", "?")) {DefaultSymbolID = "cor?"},
new SymbolicFeature("voice", new FeatureSymbol("voice+", "+"), new FeatureSymbol("voice-", "-"), new FeatureSymbol("voice?", "?")) {DefaultSymbolID = "voice?"},
new SymbolicFeature("cont", new FeatureSymbol("cont+", "+"), new FeatureSymbol("cont-", "-"), new FeatureSymbol("cont?", "?")) {DefaultSymbolID = "cont?"},
new SymbolicFeature("nas", new FeatureSymbol("nas+", "+"), new FeatureSymbol("nas-", "-"), new FeatureSymbol("nas?", "?")) {DefaultSymbolID = "nas?"},
new SymbolicFeature("str", new FeatureSymbol("str+", "+"), new FeatureSymbol("str-", "-"), new FeatureSymbol("str?", "?")) {DefaultSymbolID = "str?"},
new StringFeature("strRep")
};
WordFeatSys = new FeatureSystem
{
new SymbolicFeature("POS", new FeatureSymbol("noun"), new FeatureSymbol("verb"), new FeatureSymbol("adj"), new FeatureSymbol("adv"), new FeatureSymbol("det"))
};
Word = new FeatureSymbol("Word");
NP = new FeatureSymbol("NP");
VP = new FeatureSymbol("VP");
Seg = new FeatureSymbol("Seg");
Bdry = new FeatureSymbol("Bdry");
Allo = new FeatureSymbol("Allo");
Type = new SymbolicFeature("Type", Word, NP, VP, Seg, Bdry, Allo);
TypeFeatSys = new FeatureSystem {Type};
Characters = new Dictionary<char, FeatureStruct>
{
{'b', FeatureStruct.New(PhoneticFeatSys)
.Symbol(Seg)
.Symbol("son-")
.Symbol("syl-")
.Symbol("cons+")
.Symbol("high-")
.Symbol("ant+")
.Symbol("cor-")
.Symbol("voice+")
.Symbol("cont-")
.Symbol("nas-")
.Symbol("str-").Value},
{'d', FeatureStruct.New(PhoneticFeatSys)
.Symbol(Seg)
.Symbol("son-")
.Symbol("syl-")
.Symbol("cons+")
.Symbol("high-")
.Symbol("ant+")
.Symbol("cor+")
.Symbol("voice+")
.Symbol("cont-")
.Symbol("nas-")
.Symbol("str-").Value},
{'g', FeatureStruct.New(PhoneticFeatSys)
.Symbol(Seg)
.Symbol("son-")
.Symbol("syl-")
.Symbol("cons+")
.Symbol("high+")
.Symbol("ant-")
.Symbol("cor-")
.Symbol("voice+")
.Symbol("cont-")
.Symbol("nas-")
.Symbol("str-").Value},
{'p', FeatureStruct.New(PhoneticFeatSys)
.Symbol(Seg)
.Symbol("son-")
.Symbol("syl-")
.Symbol("cons+")
.Symbol("high-")
.Symbol("ant+")
.Symbol("cor-")
.Symbol("voice-")
.Symbol("cont-")
.Symbol("nas-")
.Symbol("str-").Value},
{'t', FeatureStruct.New(PhoneticFeatSys)
.Symbol(Seg)
.Symbol("son-")
.Symbol("syl-")
.Symbol("cons+")
.Symbol("high-")
.Symbol("ant+")
.Symbol("cor+")
.Symbol("voice-")
.Symbol("cont-")
.Symbol("nas-")
.Symbol("str-").Value},
{'q', FeatureStruct.New(PhoneticFeatSys)
.Symbol(Seg)
.Symbol("son-")
.Symbol("syl-")
.Symbol("cons+")
.Symbol("high+")
.Symbol("ant-")
.Symbol("cor-")
.Symbol("voice-")
.Symbol("cont-")
.Symbol("nas-")
.Symbol("str-").Value},
{'c', FeatureStruct.New(PhoneticFeatSys)
.Symbol(Seg)
.Symbol("son-")
.Symbol("syl-")
.Symbol("cons+")
.Symbol("high+")
.Symbol("ant-")
.Symbol("cor-")
.Symbol("voice-")
.Symbol("cont-")
.Symbol("nas-")
.Symbol("str-").Value},
{'k', FeatureStruct.New(PhoneticFeatSys)
.Symbol(Seg)
.Symbol("son-")
.Symbol("syl-")
.Symbol("cons+")
.Symbol("high+")
.Symbol("ant-")
.Symbol("cor-")
.Symbol("voice-")
.Symbol("cont-")
.Symbol("nas-")
.Symbol("str-").Value},
{'x', FeatureStruct.New(PhoneticFeatSys)
.Symbol(Seg)
.Symbol("son-")
.Symbol("syl-")
.Symbol("cons+")
.Symbol("high+")
.Symbol("ant-")
.Symbol("cor-")
.Symbol("voice-")
.Symbol("cont-")
.Symbol("nas-")
.Symbol("str-").Value},
{'j', FeatureStruct.New(PhoneticFeatSys)
.Symbol(Seg)
.Symbol("son-")
.Symbol("syl-")
.Symbol("cons+")
.Symbol("high+")
.Symbol("ant-")
.Symbol("cor+")
.Symbol("voice+")
.Symbol("cont-")
.Symbol("nas-")
.Symbol("str+").Value},
{'s', FeatureStruct.New(PhoneticFeatSys)
.Symbol(Seg)
.Symbol("son-")
.Symbol("syl-")
.Symbol("cons+")
.Symbol("high-")
.Symbol("ant+")
.Symbol("cor+")
.Symbol("voice-")
.Symbol("cont+")
.Symbol("nas-")
.Symbol("str+").Value},
{'z', FeatureStruct.New(PhoneticFeatSys)
.Symbol(Seg)
.Symbol("son-")
.Symbol("syl-")
.Symbol("cons+")
.Symbol("high-")
.Symbol("ant+")
.Symbol("cor+")
.Symbol("voice+")
.Symbol("cont+")
.Symbol("nas-")
.Symbol("str+").Value},
{'f', FeatureStruct.New(PhoneticFeatSys)
.Symbol(Seg)
.Symbol("son-")
.Symbol("syl-")
.Symbol("cons+")
.Symbol("high-")
.Symbol("ant+")
.Symbol("cor-")
.Symbol("voice-")
.Symbol("cont+")
.Symbol("nas-")
.Symbol("str+").Value},
{'v', FeatureStruct.New(PhoneticFeatSys)
.Symbol(Seg)
.Symbol("son-")
.Symbol("syl-")
.Symbol("cons+")
.Symbol("high-")
.Symbol("ant+")
.Symbol("cor-")
.Symbol("voice+")
.Symbol("cont+")
.Symbol("nas-")
.Symbol("str+").Value},
{'w', FeatureStruct.New(PhoneticFeatSys)
.Symbol(Seg)
.Symbol("son+")
.Symbol("syl-")
.Symbol("cons-")
.Symbol("high+")
.Symbol("back+")
.Symbol("front-")
.Symbol("low-")
.Symbol("rnd+")
.Symbol("ant-")
.Symbol("cor-").Value},
{'y', FeatureStruct.New(PhoneticFeatSys)
.Symbol(Seg)
.Symbol("son+")
.Symbol("syl-")
.Symbol("cons-")
.Symbol("high+")
.Symbol("back-")
.Symbol("front+")
.Symbol("low-")
.Symbol("rnd-")
.Symbol("ant-")
.Symbol("cor-").Value},
{'h', FeatureStruct.New(PhoneticFeatSys)
.Symbol(Seg)
.Symbol("son+")
.Symbol("syl-")
.Symbol("cons-")
.Symbol("high-")
.Symbol("back-")
.Symbol("front-")
.Symbol("low+")
.Symbol("ant-")
.Symbol("cor-")
.Symbol("voice-")
.Symbol("cont+")
.Symbol("nas-")
.Symbol("str-").Value},
{'r', FeatureStruct.New(PhoneticFeatSys)
.Symbol(Seg)
.Symbol("son+")
.Symbol("syl-")
.Symbol("cons+")
.Symbol("high-")
.Symbol("back-")
.Symbol("front-")
.Symbol("low-")
.Symbol("ant-")
.Symbol("cor+")
.Symbol("voice+")
.Symbol("cont+")
.Symbol("nas-")
.Symbol("str-").Value},
{'l', FeatureStruct.New(PhoneticFeatSys)
.Symbol(Seg)
.Symbol("son+")
.Symbol("syl-")
.Symbol("cons+")
.Symbol("high-")
.Symbol("back-")
.Symbol("front-")
.Symbol("low-")
.Symbol("ant+")
.Symbol("cor+")
.Symbol("voice+")
.Symbol("cont+")
.Symbol("nas-")
.Symbol("str-").Value},
{'m', FeatureStruct.New(PhoneticFeatSys)
.Symbol(Seg)
.Symbol("son+")
.Symbol("syl-")
.Symbol("cons+")
.Symbol("high-")
.Symbol("low-")
.Symbol("ant+")
.Symbol("cor-")
.Symbol("voice+")
.Symbol("cont-")
.Symbol("nas+")
.Symbol("str-").Value},
{'n', FeatureStruct.New(PhoneticFeatSys)
.Symbol(Seg)
.Symbol("son+")
.Symbol("syl-")
.Symbol("cons+")
.Symbol("high-")
.Symbol("low-")
.Symbol("ant+")
.Symbol("cor+")
.Symbol("voice+")
.Symbol("cont-")
.Symbol("nas+")
.Symbol("str-").Value},
{'N', FeatureStruct.New(PhoneticFeatSys)
.Symbol(Seg)
.Symbol("son+")
.Symbol("syl-")
.Symbol("cons+")
.Symbol("high-")
.Symbol("low-")
.Symbol("ant+")
.Symbol("cor?")
.Symbol("voice+")
.Symbol("cont-")
.Symbol("nas+")
.Symbol("str-").Value},
{'a', FeatureStruct.New(PhoneticFeatSys)
.Symbol(Seg)
.Symbol("son+")
.Symbol("syl+")
.Symbol("cons-")
.Symbol("high-")
.Symbol("back-")
.Symbol("front+")
.Symbol("low+")
.Symbol("rnd-").Value},
{'e', FeatureStruct.New(PhoneticFeatSys)
.Symbol(Seg)
.Symbol("son+")
.Symbol("syl+")
.Symbol("cons-")
.Symbol("high-")
.Symbol("back-")
.Symbol("front+")
.Symbol("low-")
.Symbol("rnd-").Value},
{'i', FeatureStruct.New(PhoneticFeatSys)
.Symbol(Seg)
.Symbol("son+")
.Symbol("syl+")
.Symbol("cons-")
.Symbol("high+")
.Symbol("back-")
.Symbol("front+")
.Symbol("low-")
.Symbol("rnd-").Value},
{'o', FeatureStruct.New(PhoneticFeatSys)
.Symbol(Seg)
.Symbol("son+")
.Symbol("syl+")
.Symbol("cons-")
.Symbol("high-")
.Symbol("back+")
.Symbol("front-")
.Symbol("low+")
.Symbol("rnd-").Value},
{'u', FeatureStruct.New(PhoneticFeatSys)
.Symbol(Seg)
.Symbol("son+")
.Symbol("syl+")
.Symbol("cons-")
.Symbol("high+")
.Symbol("back+")
.Symbol("front-")
.Symbol("low-")
.Symbol("rnd+").Value},
{'+', FeatureStruct.New(PhoneticFeatSys)
.Symbol(Bdry)
.Feature("strRep").EqualTo("+").Value},
{',', FeatureStruct.New(PhoneticFeatSys)
.Symbol(Bdry)
.Feature("strRep").EqualTo(",").Value},
{' ', FeatureStruct.New(PhoneticFeatSys)
.Symbol(Bdry)
.Feature("strRep").EqualTo(" ").Value},
{'.', FeatureStruct.New(PhoneticFeatSys)
.Symbol(Bdry)
.Feature("strRep").EqualTo(".").Value}
};
}
private static CogProject LoadProject(SpanFactory <ShapeNode> spanFactory, SegmentPool segmentPool, XDocument doc)
{
var project = new CogProject(spanFactory);
XElement root = doc.Root;
Debug.Assert(root != null);
if (root.GetDefaultNamespace() != DefaultNamespace)
{
throw new ConfigException("The specified file is not a valid Cog config file");
}
doc.Validate(Schema, (sender, args) =>
{
switch (args.Severity)
{
case XmlSeverityType.Error:
throw new ConfigException("The specified file is not a valid Cog config file", args.Exception);
}
});
segmentPool.Reset();
project.Version = (int)root.Attribute("version");
var featSys = new FeatureSystem();
XElement featSysElem = root.Element(Cog + "FeatureSystem");
Debug.Assert(featSysElem != null);
foreach (XElement featureElem in featSysElem.Elements(Cog + "Feature"))
{
var feat = new SymbolicFeature((string)featureElem.Attribute("id"), featureElem.Elements(Cog + "Value")
.Select(e => new FeatureSymbol((string)e.Attribute("id"), (string)e.Attribute("name"))))
{
Description = (string)featureElem.Attribute("name")
};
featSys.Add(feat);
}
featSys.Freeze();
project.FeatureSystem = featSys;
XElement segmentationElem = root.Element(Cog + "Segmentation");
Debug.Assert(segmentationElem != null);
XElement vowelsElem = segmentationElem.Element(Cog + "Vowels");
if (vowelsElem != null)
{
XAttribute maxLenAttr = vowelsElem.Attribute("maxLength");
if (maxLenAttr != null)
{
project.Segmenter.MaxVowelLength = (int)maxLenAttr;
}
ParseSymbols(project.FeatureSystem, vowelsElem, project.Segmenter.Vowels);
}
XElement consElem = segmentationElem.Element(Cog + "Consonants");
if (consElem != null)
{
XAttribute maxLenAttr = consElem.Attribute("maxLength");
if (maxLenAttr != null)
{
project.Segmenter.MaxConsonantLength = (int)maxLenAttr;
}
ParseSymbols(project.FeatureSystem, consElem, project.Segmenter.Consonants);
}
XElement modsElem = segmentationElem.Element(Cog + "Modifiers");
if (modsElem != null)
{
ParseSymbols(project.FeatureSystem, modsElem, project.Segmenter.Modifiers);
}
XElement bdrysElem = segmentationElem.Element(Cog + "Boundaries");
if (bdrysElem != null)
{
ParseSymbols(project.FeatureSystem, bdrysElem, project.Segmenter.Boundaries);
}
XElement tonesElem = segmentationElem.Element(Cog + "ToneLetters");
if (tonesElem != null)
{
ParseSymbols(project.FeatureSystem, tonesElem, project.Segmenter.ToneLetters);
}
XElement joinersElem = segmentationElem.Element(Cog + "Joiners");
if (joinersElem != null)
{
ParseSymbols(project.FeatureSystem, joinersElem, project.Segmenter.Joiners);
}
XElement alignersElem = root.Element(Cog + "WordAligners");
Debug.Assert(alignersElem != null);
foreach (XElement alignerElem in alignersElem.Elements(Cog + "WordAligner"))
{
LoadComponent(spanFactory, segmentPool, project, alignerElem, project.WordAligners);
}
XElement cognateIdentifiersElem = root.Element(Cog + "CognateIdentifiers");
Debug.Assert(cognateIdentifiersElem != null);
foreach (XElement cognateIdentifierElem in cognateIdentifiersElem.Elements(Cog + "CognateIdentifier"))
{
LoadComponent(spanFactory, segmentPool, project, cognateIdentifierElem, project.CognateIdentifiers);
}
var meanings = new Dictionary <string, Meaning>();
XElement meaningsElem = root.Element(Cog + "Meanings");
Debug.Assert(meaningsElem != null);
foreach (XElement meaningElem in meaningsElem.Elements(Cog + "Meaning"))
{
var meaning = new Meaning((string)meaningElem.Attribute("gloss"), (string)meaningElem.Attribute("category"));
meanings[(string)meaningElem.Attribute("id")] = meaning;
project.Meanings.Add(meaning);
}
XElement varietiesElem = root.Element(Cog + "Varieties");
Debug.Assert(varietiesElem != null);
foreach (XElement varietyElem in varietiesElem.Elements(Cog + "Variety"))
{
var variety = new Variety((string)varietyElem.Attribute("name"));
XElement wordsElem = varietyElem.Element(Cog + "Words");
if (wordsElem != null)
{
foreach (XElement wordElem in wordsElem.Elements(Cog + "Word"))
{
Meaning meaning;
if (meanings.TryGetValue((string)wordElem.Attribute("meaning"), out meaning))
{
var strRep = ((string)wordElem).Trim();
var stemIndex = (int?)wordElem.Attribute("stemIndex") ?? 0;
var stemLen = (int?)wordElem.Attribute("stemLength") ?? strRep.Length - stemIndex;
variety.Words.Add(new Word(strRep, stemIndex, stemLen, meaning));
}
}
}
XElement affixesElem = varietyElem.Element(Cog + "Affixes");
if (affixesElem != null)
{
foreach (XElement affixElem in affixesElem.Elements(Cog + "Affix"))
{
var type = AffixType.Prefix;
switch ((string)affixElem.Attribute("type"))
{
case "prefix":
type = AffixType.Prefix;
break;
case "suffix":
type = AffixType.Suffix;
break;
}
var affixStr = ((string)affixElem).Trim();
variety.Affixes.Add(new Affix(affixStr, type, (string)affixElem.Attribute("category")));
}
}
XElement regionsElem = varietyElem.Element(Cog + "Regions");
if (regionsElem != null)
{
foreach (XElement regionElem in regionsElem.Elements(Cog + "Region"))
{
var region = new GeographicRegion {
Description = (string)regionElem.Element(Cog + "Description")
};
foreach (XElement coordinateElem in regionElem.Elements(Cog + "Coordinates").Elements(Cog + "Coordinate"))
{
var latitude = (double)coordinateElem.Element(Cog + "Latitude");
var longitude = (double)coordinateElem.Element(Cog + "Longitude");
region.Coordinates.Add(new GeographicCoordinate(latitude, longitude));
}
variety.Regions.Add(region);
}
}
project.Varieties.Add(variety);
}
XElement projectProcessorsElem = root.Element(Cog + "ProjectProcessors");
Debug.Assert(projectProcessorsElem != null);
foreach (XElement projectProcessorElem in projectProcessorsElem.Elements(Cog + "ProjectProcessor"))
{
LoadComponent(spanFactory, segmentPool, project, projectProcessorElem, project.ProjectProcessors);
}
XElement varietyProcessorsElem = root.Element(Cog + "VarietyProcessors");
Debug.Assert(varietyProcessorsElem != null);
foreach (XElement varietyProcessorElem in varietyProcessorsElem.Elements(Cog + "VarietyProcessor"))
{
LoadComponent(spanFactory, segmentPool, project, varietyProcessorElem, project.VarietyProcessors);
}
XElement varietyPairProcessorsElem = root.Element(Cog + "VarietyPairProcessors");
Debug.Assert(varietyPairProcessorsElem != null);
foreach (XElement varietyPairProcessorElem in varietyPairProcessorsElem.Elements(Cog + "VarietyPairProcessor"))
{
LoadComponent(spanFactory, segmentPool, project, varietyPairProcessorElem, project.VarietyPairProcessors);
}
return(project);
}
private void LoadLanguage(XElement langElem)
{
_language = new Language { Name = (string) langElem.Element("Name") };
IEnumerable<FeatureSymbol> posSymbols = langElem.Elements("PartsOfSpeech").Elements("PartOfSpeech")
.Select(e => new FeatureSymbol((string) e.Attribute("id"), (string) e.Element("Name")));
_posFeature = _language.SyntacticFeatureSystem.AddPartsOfSpeech(posSymbols);
XElement phonFeatSysElem = langElem.Elements("PhonologicalFeatureSystem").SingleOrDefault(IsActive);
if (phonFeatSysElem != null)
LoadPhonologicalFeatureSystem(phonFeatSysElem);
_language.PhonologicalFeatureSystem.Freeze();
XElement headFeatsElem = langElem.Element("HeadFeatures");
if (headFeatsElem != null)
{
_headFeature = _language.SyntacticFeatureSystem.AddHeadFeature();
LoadSyntacticFeatureSystem(headFeatsElem, SyntacticFeatureType.Head);
}
XElement footFeatsElem = langElem.Element("FootFeatures");
if (footFeatsElem != null)
{
_footFeature = _language.SyntacticFeatureSystem.AddFootFeature();
LoadSyntacticFeatureSystem(footFeatsElem, SyntacticFeatureType.Foot);
}
_language.SyntacticFeatureSystem.Freeze();
foreach (XElement mfElem in langElem.Elements("MorphologicalPhonologicalRuleFeatures").Elements("MorphologicalPhonologicalRuleFeature").Where(IsActive))
{
var mprFeature = new MprFeature {Name = (string) mfElem};
_language.MprFeatures.Add(mprFeature);
_mprFeatures[(string) mfElem.Attribute("id")] = mprFeature;
}
foreach (XElement mprFeatGroupElem in langElem.Elements("MorphologicalPhonologicalRuleFeatures").Elements("MorphologicalPhonologicalRuleFeatureGroup").Where(IsActive))
LoadMprFeatGroup(mprFeatGroupElem);
foreach (XElement stemNameElem in langElem.Elements("StemNames").Elements("StemName"))
LoadStemName(stemNameElem);
foreach (XElement charDefTableElem in langElem.Elements("CharacterDefinitionTable").Where(IsActive))
LoadCharacterDefinitionTable(charDefTableElem);
foreach (XElement natClassElem in langElem.Elements("NaturalClasses").Elements().Where(IsActive))
LoadNaturalClass(natClassElem);
foreach (XElement familyElem in langElem.Elements("Families").Elements("Family").Where(IsActive))
{
var family = new LexFamily {Name = (string) familyElem};
_language.Families.Add(family);
_families[(string) familyElem.Attribute("id")] = family;
}
foreach (XElement pruleElem in langElem.Elements("PhonologicalRuleDefinitions").Elements().Where(IsActive))
LoadPhonologicalRule(pruleElem);
foreach (XElement stratumElem in langElem.Elements("Strata").Elements("Stratum").Where(IsActive))
LoadStratum(stratumElem);
foreach (XElement coOccurElem in langElem.Elements("MorphemeCoOccurrenceRules").Elements("MorphemeCoOccurrenceRule").Where(IsActive))
LoadMorphemeCoOccurrenceRule(coOccurElem);
foreach (XElement coOccurElem in langElem.Elements("AllomorphCoOccurrenceRules").Elements("AllomorphCoOccurrenceRule").Where(IsActive))
LoadAllomorphCoOccurrenceRule(coOccurElem);
}
public void Migrate(SegmentPool segmentPool, CogProject project)
{
if (project.FeatureSystem.ContainsFeature("airstream"))
{
return;
}
var pulmonic = new FeatureSymbol("pulmonic");
var ejective = new FeatureSymbol("ejective");
var implosive = new FeatureSymbol("implosive");
var click = new FeatureSymbol("click");
var airstream = new SymbolicFeature("airstream",
pulmonic,
ejective,
implosive,
click)
{
Description = "Airstream"
};
project.FeatureSystem.Add(airstream);
AddValue(project.Segmenter.Modifiers, "ʼ", ejective);
AddValue(project.Segmenter.Consonants, "ɓ", implosive);
AddValue(project.Segmenter.Consonants, "ɗ", implosive);
AddValue(project.Segmenter.Consonants, "ʄ", implosive);
AddValue(project.Segmenter.Consonants, "ɠ", implosive);
AddValue(project.Segmenter.Consonants, "ʛ", implosive);
FeatureSymbol affricate;
if (project.FeatureSystem.TryGetSymbol("affricate", out affricate))
{
project.Segmenter.Consonants.AddSymbolBasedOn("ʘ", "p", affricate, click);
project.Segmenter.Consonants.AddSymbolBasedOn("ǀ", "θ", affricate, click);
project.Segmenter.Consonants.AddSymbolBasedOn("ǁ", "ɬ", affricate, click);
}
project.Segmenter.Consonants.AddSymbolBasedOn("ǃ", "t", click);
project.Segmenter.Consonants.AddSymbolBasedOn("ǂ", "c", click);
foreach (Symbol symbol in project.Segmenter.Vowels.ToArray())
{
AddValue(project.Segmenter.Vowels, symbol, pulmonic);
}
foreach (Symbol symbol in project.Segmenter.Consonants.Where(s => !s.FeatureStruct.ContainsFeature("airstream")).ToArray())
{
AddValue(project.Segmenter.Consonants, symbol, pulmonic);
}
foreach (KeyValuePair <string, IWordAligner> aligner in project.WordAligners.Where(kvp => kvp.Value is Aline).ToArray())
{
var aline = (Aline)aligner.Value;
Dictionary <SymbolicFeature, int> featWeights = aline.FeatureWeights.ToDictionary(kvp => kvp.Key, kvp => kvp.Value);
featWeights[airstream] = 5;
Dictionary <FeatureSymbol, int> valueMetrics = aline.ValueMetrics.ToDictionary(kvp => kvp.Key, kvp => kvp.Value);
valueMetrics[pulmonic] = 100;
valueMetrics[ejective] = 66;
valueMetrics[implosive] = 33;
valueMetrics[click] = 0;
project.WordAligners[aligner.Key] = new Aline(segmentPool, aline.RelevantVowelFeatures, aline.RelevantConsonantFeatures.Concat(airstream),
featWeights, valueMetrics, aline.Settings);
}
}
static CogFeatureSystem()
{
AnchorType = new FeatureSymbol(Guid.NewGuid().ToString())
{
Description = "anchor"
};
VowelType = new FeatureSymbol(Guid.NewGuid().ToString())
{
Description = "vowel"
};
ConsonantType = new FeatureSymbol(Guid.NewGuid().ToString())
{
Description = "consonant"
};
ToneLetterType = new FeatureSymbol(Guid.NewGuid().ToString())
{
Description = "toneLetter"
};
BoundaryType = new FeatureSymbol(Guid.NewGuid().ToString())
{
Description = "boundary"
};
SyllableType = new FeatureSymbol(Guid.NewGuid().ToString())
{
Description = "syllable"
};
StemType = new FeatureSymbol(Guid.NewGuid().ToString())
{
Description = "stem"
};
PrefixType = new FeatureSymbol(Guid.NewGuid().ToString())
{
Description = "prefix"
};
SuffixType = new FeatureSymbol(Guid.NewGuid().ToString())
{
Description = "suffix"
};
Type = new SymbolicFeature(Guid.NewGuid().ToString(), AnchorType, VowelType, ConsonantType, ToneLetterType, BoundaryType, SyllableType, StemType, PrefixType, SuffixType)
{
Description = "Type"
};
Complex = new FeatureSymbol(Guid.NewGuid().ToString())
{
Description = "complex"
};
Simple = new FeatureSymbol(Guid.NewGuid().ToString())
{
Description = "simple"
};
SegmentType = new SymbolicFeature(Guid.NewGuid().ToString(), Complex, Simple)
{
Description = "SegmentType"
};
OriginalStrRep = new StringFeature(Guid.NewGuid().ToString())
{
Description = "OriginalStrRep"
};
StrRep = new StringFeature(Guid.NewGuid().ToString())
{
Description = "StrRep"
};
First = new ComplexFeature(Guid.NewGuid().ToString())
{
Description = "First"
};
Onset = new FeatureSymbol(Guid.NewGuid().ToString())
{
Description = "onset"
};
Nucleus = new FeatureSymbol(Guid.NewGuid().ToString())
{
Description = "nucleus"
};
Coda = new FeatureSymbol(Guid.NewGuid().ToString())
{
Description = "coda"
};
SyllablePosition = new SymbolicFeature(Guid.NewGuid().ToString(), Onset, Nucleus, Coda)
{
Description = "SyllablePosition"
};
Instance = new CogFeatureSystem();
}
private Feature LoadFeature(XElement featElem)
{
var id = (string) featElem.Attribute("id");
var name = (string) featElem.Element("Name");
switch (featElem.Name.LocalName)
{
case "SymbolicFeature":
IEnumerable<FeatureSymbol> symbols = featElem.Elements("Symbols").Elements("Symbol").Select(e => new FeatureSymbol((string) e.Attribute("id"), (string) e));
var feature = new SymbolicFeature(id, symbols) { Description = name };
var defValId = (string) featElem.Attribute("defaultSymbol");
if (!string.IsNullOrEmpty(defValId))
feature.DefaultSymbolID = defValId;
return feature;
case "ComplexFeature":
return new ComplexFeature(id) {Description = name};
}
return null;
}
