public Word(string original, PosTaggerOutputType type, int startIndex = -1)
{
Original = original;
Type = type;
StartIndex = startIndex;
ErrorCodes = "";
}
public static string ToText(this PosTaggerOutputType posTaggerOutputType)
{
switch (posTaggerOutputType)
{
case PosTaggerOutputType.Article: return("Article");
case PosTaggerOutputType.Adjective: return("Adj");
case PosTaggerOutputType.AdjectivePronoun: return("AdjPron");
case PosTaggerOutputType.Adverb: return("Adv");
case PosTaggerOutputType.AdverbialParticiple: return("AdvPart");
case PosTaggerOutputType.AdverbialPronoun: return("AdvPron");
case PosTaggerOutputType.AuxiliaryVerb: return("AuxVerb");
case PosTaggerOutputType.Conjunction: return("Conj");
case PosTaggerOutputType.Gerund: return("Gerund");
case PosTaggerOutputType.Infinitive: return("Inf");
case PosTaggerOutputType.Interjection: return("Intr");
case PosTaggerOutputType.ModalVerb: return("ModalVerb");
case PosTaggerOutputType.Noun: return("Noun");
case PosTaggerOutputType.Numeral: return("Num");
case PosTaggerOutputType.PastParticiple: return("PastParticiple");
case PosTaggerOutputType.Participle: return("Part");
case PosTaggerOutputType.Particle: return("Pr");
case PosTaggerOutputType.PossessivePronoun: return("PosPron");
case PosTaggerOutputType.Predicate: return("Pred");
case PosTaggerOutputType.Preposition: return("Prep");
case PosTaggerOutputType.Pronoun: return("Pron");
case PosTaggerOutputType.Punctuation: return("Punct");
case PosTaggerOutputType.Verb: return("Verb");
default: //case PosTaggerOutputType.Other:
return("Other");
}
}
public static byte ToCrfByte(this PosTaggerOutputType posTaggerOutputType)
{
switch (posTaggerOutputType)
{
case PosTaggerOutputType.Adjective: return((byte)'J');
case PosTaggerOutputType.AdjectivePronoun: return((byte)'R');
case PosTaggerOutputType.Adverb: return((byte)'D');
case PosTaggerOutputType.AdverbialParticiple: return((byte)'X');
case PosTaggerOutputType.AdverbialPronoun: return((byte)'H');
case PosTaggerOutputType.Article: return((byte)'A');
case PosTaggerOutputType.AuxiliaryVerb: return((byte)'G');
case PosTaggerOutputType.Conjunction: return((byte)'C');
case PosTaggerOutputType.Gerund: return((byte)'L');
case PosTaggerOutputType.Infinitive: return((byte)'F');
case PosTaggerOutputType.Interjection: return((byte)'I');
case PosTaggerOutputType.ModalVerb: return((byte)'K');
case PosTaggerOutputType.Noun: return((byte)'N');
case PosTaggerOutputType.Numeral: return((byte)'M');
case PosTaggerOutputType.PastParticiple: return((byte)'B');
case PosTaggerOutputType.Participle: return((byte)'Z');
case PosTaggerOutputType.Particle: return((byte)'W');
case PosTaggerOutputType.PossessivePronoun: return((byte)'S');
case PosTaggerOutputType.Preposition: return((byte)'E');
case PosTaggerOutputType.Pronoun: return((byte)'Y');
case PosTaggerOutputType.Punctuation: return((byte)'T');
case PosTaggerOutputType.Verb: return((byte)'V');
default: //case PosTaggerOutputType.Other:
return((byte)'O');
}
}
internal static PartOfSpeechEnum?ToPartOfSpeech(PosTaggerOutputType posTaggerOutputType)
{
switch (posTaggerOutputType)
{
case PosTaggerOutputType.Adjective:
/*case PosTaggerOutputType.AdjectivePronoun :*/ return(PartOfSpeechEnum.Adjective);
case PosTaggerOutputType.Adverb:
/*case PosTaggerOutputType.AdverbialPronoun :*/ return(PartOfSpeechEnum.Adverb);
case PosTaggerOutputType.Article: return(PartOfSpeechEnum.Article);
case PosTaggerOutputType.Conjunction: return(PartOfSpeechEnum.Conjunction);
case PosTaggerOutputType.Interjection: return(PartOfSpeechEnum.Interjection);
case PosTaggerOutputType.Noun: return(PartOfSpeechEnum.Noun);
case PosTaggerOutputType.Numeral: return(PartOfSpeechEnum.Numeral);
case PosTaggerOutputType.Other: return(PartOfSpeechEnum.Other);
case PosTaggerOutputType.Particle: return(PartOfSpeechEnum.Particle);
case PosTaggerOutputType.Predicate: return(PartOfSpeechEnum.Predicate);
case PosTaggerOutputType.Preposition: return(PartOfSpeechEnum.Preposition);
case PosTaggerOutputType.Pronoun:
case PosTaggerOutputType.PossessivePronoun:
case PosTaggerOutputType.AdjectivePronoun:
case PosTaggerOutputType.AdverbialPronoun: return(PartOfSpeechEnum.Pronoun);
case PosTaggerOutputType.Verb:
case PosTaggerOutputType.Infinitive:
case PosTaggerOutputType.AdverbialParticiple:
case PosTaggerOutputType.AuxiliaryVerb:
case PosTaggerOutputType.Participle: return(PartOfSpeechEnum.Verb);
//default:
//throw (new ArgumentException(posTaggerOutputType.ToString()));
}
return(null);
}
internal static PartOfSpeechEnum?ToPartOfSpeech(PosTaggerOutputType posTaggerOutputType)
{
switch (posTaggerOutputType)
{
case PosTaggerOutputType.Adjective:
return(PartOfSpeechEnum.Adjective);
case PosTaggerOutputType.Adverb:
return(PartOfSpeechEnum.Adverb);
case PosTaggerOutputType.Article: return(PartOfSpeechEnum.Article);
case PosTaggerOutputType.Conjunction: return(PartOfSpeechEnum.Conjunction);
case PosTaggerOutputType.Interjection: return(PartOfSpeechEnum.Interjection);
case PosTaggerOutputType.Noun: return(PartOfSpeechEnum.Noun);
case PosTaggerOutputType.Numeral: return(PartOfSpeechEnum.Numeral);
case PosTaggerOutputType.Other: return(PartOfSpeechEnum.Other);
case PosTaggerOutputType.Particle: return(PartOfSpeechEnum.Particle);
case PosTaggerOutputType.Predicate: return(PartOfSpeechEnum.Predicate);
case PosTaggerOutputType.Preposition: return(PartOfSpeechEnum.Preposition);
case PosTaggerOutputType.Pronoun:
case PosTaggerOutputType.PossessivePronoun:
case PosTaggerOutputType.AdjectivePronoun:
case PosTaggerOutputType.AdverbialPronoun: return(PartOfSpeechEnum.Pronoun);
case PosTaggerOutputType.Verb:
case PosTaggerOutputType.Infinitive:
case PosTaggerOutputType.AdverbialParticiple:
case PosTaggerOutputType.AuxiliaryVerb:
case PosTaggerOutputType.Participle: return(PartOfSpeechEnum.Verb);
}
return(null);
}
public static char ToCrfChar(this PosTaggerOutputType posTaggerOutputType)
{
return((char)posTaggerOutputType.ToCrfByte());
}
unsafe private List <string> GetPosTaggerAttributes4ModelBuilder(
int wordIndex, int wordsCount, PosTaggerOutputType posTaggerOutputType)
{
var wordsCount_Minus1 = wordsCount - 1;
_Result4ModelBuilder.Clear();
_Result4ModelBuilder.Add(/*_Words[ wordIndex ].*/ posTaggerOutputType.ToCrfChar().ToString());
var ngrams = _CrfTemplateFile.GetCRFNgramsWhichCanTemplateBeApplied(wordIndex, wordsCount);
for (int i = 0, ngramsLength = ngrams.Length; i < ngramsLength; i++)
{
var ngram = ngrams[i];
_AttributeBufferPtr = ngram.CopyAttributesHeaderChars(_AttributeBufferPtrBase);
#region [.build attr-values.]
switch (ngram.CRFAttributesLength)
{
case 1:
#region
{
AppendAttrValue(wordIndex, ngram.CRFAttribute_0);
}
#endregion
break;
case 2:
#region
{
AppendAttrValue(wordIndex, ngram.CRFAttribute_0); *(_AttributeBufferPtr++) = VERTICAL_SLASH;
AppendAttrValue(wordIndex, ngram.CRFAttribute_1);
}
#endregion
break;
case 3:
#region
{
AppendAttrValue(wordIndex, ngram.CRFAttribute_0); *(_AttributeBufferPtr++) = VERTICAL_SLASH;
AppendAttrValue(wordIndex, ngram.CRFAttribute_1); *(_AttributeBufferPtr++) = VERTICAL_SLASH;
AppendAttrValue(wordIndex, ngram.CRFAttribute_2);
}
#endregion
break;
default:
#region
{
for (var j = 0; j < ngram.CRFAttributesLength; j++)
{
AppendAttrValue(wordIndex, ngram.CRFAttributes[j]); *(_AttributeBufferPtr++) = VERTICAL_SLASH;
}
// Удалить последний '|'
_AttributeBufferPtr--;
}
#endregion
break;
}
#endregion
var crfValue = new string( _AttributeBufferPtrBase, 0, (int)(_AttributeBufferPtr - _AttributeBufferPtrBase));
_Result4ModelBuilder.Add(crfValue);
}
if (wordIndex == 0)
{
_Result4ModelBuilder.Add(xlat_Unsafe.BEGIN_OF_SENTENCE);
}
else
if (wordIndex == wordsCount_Minus1)
{
_Result4ModelBuilder.Add(xlat_Unsafe.END_OF_SENTENCE);
}
return(_Result4ModelBuilder);
}
public Word(string original, PosTaggerOutputType type)
{
Original = original;
Type = type;
Errors = new List <Error>();
}
