public WordForms_t GetWordFormsByPartOfSpeech(string word, PartOfSpeechEnum partOfSpeechFilter)
{
var result = new WordForms_t(word);
var wordUpper = StringsHelper.ToUpperInvariant(word);
if (_MorphoModel.GetWordForms(wordUpper, _WordForms))
{
FillUniqueWordFormsDictionary(partOfSpeechFilter);
#region [.fill word-forms list.]
_WordForms.Clear();
foreach (var p in _UniqueWordFormsDictionary)
{
var form = p.Key;
var partOfSpeech = p.Value;
var wf = new WordForm_t(form, partOfSpeech);
_WordForms.Add(wf);
}
result.Forms = _WordForms;
#endregion
}
return(result);
}
private static PosTaggerOutputType ToPosTaggerOutputType(PartOfSpeechEnum singlePartOfSpeech)
{
switch (singlePartOfSpeech)
{
case PartOfSpeechEnum.Adjective: return(PosTaggerOutputType.Adjective);
case PartOfSpeechEnum.Adverb: return(PosTaggerOutputType.Adverb);
case PartOfSpeechEnum.Article: return(PosTaggerOutputType.Article);
case PartOfSpeechEnum.Conjunction: return(PosTaggerOutputType.Conjunction);
case PartOfSpeechEnum.Interjection: return(PosTaggerOutputType.Interjection);
case PartOfSpeechEnum.Noun: return(PosTaggerOutputType.Noun);
case PartOfSpeechEnum.Numeral: return(PosTaggerOutputType.Numeral);
case PartOfSpeechEnum.Other: return(PosTaggerOutputType.Other);
case PartOfSpeechEnum.Particle: return(PosTaggerOutputType.Particle);
case PartOfSpeechEnum.Predicate: return(PosTaggerOutputType.Predicate);
case PartOfSpeechEnum.Preposition: return(PosTaggerOutputType.Preposition);
case PartOfSpeechEnum.Pronoun: return(PosTaggerOutputType.Pronoun);
case PartOfSpeechEnum.Verb: return(PosTaggerOutputType.Verb);
default:
throw new ArgumentException(singlePartOfSpeech.ToString());
}
}
public BaseMorphoFormNative(char *_base, MorphoTypeNative morphoType)
{
_Base = _base;
_PartOfSpeech = morphoType.PartOfSpeech;
Debug.Assert(morphoType.HasMorphoForms, "morphoType.MorphoForms.Length <= 0");
_MorphoFormEndings = morphoType.MorphoFormEndings;
#region commented
/*
* for ( int i = 0, len = morphoType.MorphoForms.Length; i < len; i++ )
* {
* tempBufferHS.Add( (IntPtr) morphoType.MorphoForms[ i ].Ending );
* }
* _MorphoFormEndings = new char*[ tempBufferHS.Count ];
* fixed ( char** morphoFormEndingsBase = _MorphoFormEndings )
* {
* var morphoFormEndings_ptr = morphoFormEndingsBase;
* foreach ( var intptr in tempBufferHS )
* {
*(morphoFormEndings_ptr++) = (char*) intptr;
* }
* }
* tempBufferHS.Clear();
*/
#endregion
//_GlobalCount++;
//_GlobalHashsetBase.Add( _Base );
//_GlobalHashsetNormalForm.Add( _NormalForm );
}
internal MorphoTypeNative(PartOfSpeechBase partOfSpeechBase)
{
_morphoAttributeGroup = partOfSpeechBase.MorphoAttributeGroup;
_partOfSpeech = partOfSpeechBase.PartOfSpeech;
_morphoFormEndings = EMPTY_ENDINGS;
_morphoFormEndingUpperAndMorphoAttributes = EMPTY_MFUEMA;
}
/// получение уникальных форм
/// pForms - все формы
/// uniqueForms [out] - уникальные формы
/// pos - часть речи
/// result - общее число уникальных форм
private void FillUniqueWordFormsDictionary(PartOfSpeechEnum partOfSpeechFilter)
{
_uniqueWordFormsDictionary.Clear();
for (int i = 0, len = _wordForms.Count; i < len; i++)
{
var wordForm = _wordForms[i];
PartOfSpeechEnum partOfSpeechForm = wordForm.PartOfSpeech;
if ((partOfSpeechForm & partOfSpeechFilter) != partOfSpeechFilter)
{
continue;
}
var wordFormString = wordForm.Form;
if (_uniqueWordFormsDictionary.TryGetValue(wordFormString, out PartOfSpeechEnum partOfSpeechExists))
{
if ((partOfSpeechExists & partOfSpeechForm) != partOfSpeechForm)
{
partOfSpeechExists |= partOfSpeechForm;
_uniqueWordFormsDictionary[wordFormString] = partOfSpeechExists;
}
}
else
{
_uniqueWordFormsDictionary.Add(wordFormString, partOfSpeechForm);
}
}
}
/// Получение части речи по ее названию
public PartOfSpeechBase GetPartOfSpeech(PartOfSpeechEnum partOfSpeech)
{
if (_Dictionary.TryGetValue(partOfSpeech, out PartOfSpeechBase value))
{
return(value);
}
return(null);
}
public BaseMorphoFormNative(char *_base, MorphoTypeNative morphoType)
{
this._base = _base;
_partOfSpeech = morphoType.PartOfSpeech;
Debug.Assert(morphoType.HasMorphoForms, "morphoType.MorphoForms.Length <= 0");
_morphoFormEndings = morphoType.MorphoFormEndings;
}
private WordFormMorphology_t[] TryGetByPosTaggerOutputType(ref WordMorphology_t wordMorphology
, PartOfSpeechEnum filterPartOfSpeech)
{
if ((wordMorphology.PartOfSpeech & filterPartOfSpeech) == filterPartOfSpeech)
{
var morphologies = TryGetByPosTaggerOutputType(wordMorphology.WordFormMorphologies, filterPartOfSpeech);
return(morphologies);
}
return(null);
}
private WordFormMorphology[] TryGetByPosTaggerOutputType(ref WordMorphology wordMorphology,
PartOfSpeechEnum filterPartOfSpeech_1, PartOfSpeechEnum filterPartOfSpeech_2)
{
if ((wordMorphology.PartOfSpeech & filterPartOfSpeech_1) == filterPartOfSpeech_1)
{
var morphologies = TryGetByPosTaggerOutputType(wordMorphology.WordFormMorphologies, filterPartOfSpeech_1);
if (morphologies != null)
{
return(morphologies);
}
}
if ((wordMorphology.PartOfSpeech & filterPartOfSpeech_2) == filterPartOfSpeech_2)
{
var morphologies = TryGetByPosTaggerOutputType(wordMorphology.WordFormMorphologies, filterPartOfSpeech_2);
return(morphologies);
}
return(null);
}
public WordForms GetWordFormsByPartOfSpeech_NoToUpper(string wordUpper, PartOfSpeechEnum partOfSpeechFilter)
{
var result = new WordForms(wordUpper);
if (_morphoModel.GetWordForms(wordUpper, _wordForms))
{
FillUniqueWordFormsDictionary(partOfSpeechFilter);
_wordForms.Clear();
foreach (var p in _uniqueWordFormsDictionary)
{
var form = p.Key;
var partOfSpeech = p.Value;
var wf = new WordForm(form, partOfSpeech);
_wordForms.Add(wf);
}
result.Forms = _wordForms;
}
return(result);
}
private WordFormMorphology_t[] TryGetByPosTaggerOutputType(List <WordFormMorphology_t> wordFormMorphologies
, PartOfSpeechEnum filterPartOfSpeech)
{
_wordFormMorphologies_Buffer.Clear();
for (int i = 0, len = wordFormMorphologies.Count; i < len; i++)
{
var wordFormMorphology = wordFormMorphologies[i];
if ((wordFormMorphology.PartOfSpeech & filterPartOfSpeech) == filterPartOfSpeech)
{
//if ( FilterByLetterCase( ref wordFormMorphology, _WordFirstCharIsUpper ) )
//{
_wordFormMorphologies_Buffer.Add(wordFormMorphology);
//}
}
}
if (_wordFormMorphologies_Buffer.Count != 0)
{
var result = _wordFormMorphologies_Buffer.ToArray(); //FilterByLetterCase( _FilteredWordFormMorphologies ); //
return(result);
}
return(null);
}
private WordFormMorphology[] TryGetByPosTaggerOutputType(List <WordFormMorphology> wordFormMorphologies, PartOfSpeechEnum filterPartOfSpeech)
{
_wordFormMorphologies_Buffer.Clear();
for (int i = 0, len = wordFormMorphologies.Count; i < len; i++)
{
var wordFormMorphology = wordFormMorphologies[i];
if ((wordFormMorphology.PartOfSpeech & filterPartOfSpeech) == filterPartOfSpeech)
{
_wordFormMorphologies_Buffer.Add(wordFormMorphology);
}
}
if (_wordFormMorphologies_Buffer.Count != 0)
{
var result = _wordFormMorphologies_Buffer.ToArray();
return(result);
}
return(null);
}
public WordFormMorphology(PartOfSpeechEnum partOfSpeech)
: this()
{
PartOfSpeech = partOfSpeech;
}
public WordForm(string form, PartOfSpeechEnum partOfSpeech)
{
Form = form;
PartOfSpeech = partOfSpeech;
}
public WordFormMorphology(PartOfSpeechEnum partOfSpeech, MorphoAttributeEnum morphoAttribute)
: this()
{
PartOfSpeech = partOfSpeech;
MorphoAttribute = morphoAttribute;
}
/*
* морфоанализатор::{PartOfSpeechEnum} PoS-tagger::{PosTaggerOutputType}
* PartOfSpeechEnum.Adjective PosTaggerOutputType.Adjective
* PosTaggerOutputType.AdjectivePronoun
* PartOfSpeechEnum.Adverb PosTaggerOutputType.Adverb
* PosTaggerOutputType.AdverbialPronoun
* PartOfSpeechEnum.Article PosTaggerOutputType.Article
* PartOfSpeechEnum.Conjunction PosTaggerOutputType.Conjunction
* PartOfSpeechEnum.Interjection PosTaggerOutputType.Interjection
* PartOfSpeechEnum.Noun PosTaggerOutputType.Noun
* PartOfSpeechEnum.Numeral PosTaggerOutputType.Numeral
* PartOfSpeechEnum.Other PosTaggerOutputType.Other
* PartOfSpeechEnum.Particle PosTaggerOutputType.Particle
* PartOfSpeechEnum.Predicate PosTaggerOutputType.Predicate
* PartOfSpeechEnum.Preposition PosTaggerOutputType.Preposition
* PartOfSpeechEnum.Pronoun PosTaggerOutputType.Pronoun
* PosTaggerOutputType.PossessivePronoun
* PosTaggerOutputType.AdjectivePronoun
* PosTaggerOutputType.AdverbialPronoun
* PartOfSpeechEnum.Verb PosTaggerOutputType.Verb
* PosTaggerOutputType.Infinitive
* PosTaggerOutputType.AdverbialParticiple
* PosTaggerOutputType.AuxiliaryVerb
* PosTaggerOutputType.Participle
* - PosTaggerOutputType.Punctuation
*/
#endregion
private static void CorrectPosTaggerOutputType(Word word, PartOfSpeechEnum singlePartOfSpeech)
{
switch (singlePartOfSpeech)
{
case PartOfSpeechEnum.Adjective:
switch (word.posTaggerOutputType)
{
case PosTaggerOutputType.Adjective:
case PosTaggerOutputType.AdjectivePronoun:
break;
default:
word.posTaggerOutputType = PosTaggerOutputType.Adjective;
break;
}
break;
case PartOfSpeechEnum.Adverb:
switch (word.posTaggerOutputType)
{
case PosTaggerOutputType.Adverb:
case PosTaggerOutputType.AdverbialPronoun:
break;
default:
word.posTaggerOutputType = PosTaggerOutputType.Adverb;
break;
}
break;
case PartOfSpeechEnum.Article: word.posTaggerOutputType = PosTaggerOutputType.Article; break;
case PartOfSpeechEnum.Conjunction: word.posTaggerOutputType = PosTaggerOutputType.Conjunction; break;
case PartOfSpeechEnum.Interjection: word.posTaggerOutputType = PosTaggerOutputType.Interjection; break;
case PartOfSpeechEnum.Noun: word.posTaggerOutputType = PosTaggerOutputType.Noun; break;
case PartOfSpeechEnum.Numeral: word.posTaggerOutputType = PosTaggerOutputType.Numeral; break;
case PartOfSpeechEnum.Other: word.posTaggerOutputType = PosTaggerOutputType.Other; break;
case PartOfSpeechEnum.Particle: word.posTaggerOutputType = PosTaggerOutputType.Particle; break;
case PartOfSpeechEnum.Predicate: word.posTaggerOutputType = PosTaggerOutputType.Predicate; break;
case PartOfSpeechEnum.Preposition: word.posTaggerOutputType = PosTaggerOutputType.Preposition; break;
case PartOfSpeechEnum.Pronoun:
switch (word.posTaggerOutputType)
{
case PosTaggerOutputType.Pronoun:
case PosTaggerOutputType.PossessivePronoun:
case PosTaggerOutputType.AdjectivePronoun:
case PosTaggerOutputType.AdverbialPronoun:
break;
default:
word.posTaggerOutputType = PosTaggerOutputType.Pronoun;
break;
}
break;
case PartOfSpeechEnum.Verb:
switch (word.posTaggerOutputType)
{
case PosTaggerOutputType.Verb:
case PosTaggerOutputType.Infinitive:
case PosTaggerOutputType.AdverbialParticiple:
case PosTaggerOutputType.AuxiliaryVerb:
case PosTaggerOutputType.Participle:
break;
default:
word.posTaggerOutputType = PosTaggerOutputType.Verb;
break;
}
break;
default:
throw new ArgumentException(singlePartOfSpeech.ToString());
}
}
public IntPtr this[PartOfSpeechEnum partOfSpeech]
{
get { return(_Dictionary[partOfSpeech]); }
}
internal MorphoType(PartOfSpeechBase partOfSpeechBase)
{
_MorphoForms = EMPTY;
_MorphoAttributeGroup = partOfSpeechBase.MorphoAttributeGroup;
_PartOfSpeech = partOfSpeechBase.PartOfSpeech;
}
public WordFormMorphology_t(PartOfSpeechEnum partOfSpeech) : this() => PartOfSpeech = partOfSpeech;
