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);
}
internal MorphoForm(string ending, List <MorphoAttributePair> morphoAttributePair)
{
Ending = string.Intern(ending);
EndingUpper = string.Intern(StringsHelper.ToUpperInvariant(Ending));
if (morphoAttributePair.Count != 0)
{
MorphoAttributePairs = morphoAttributePair.ToArray();
}
else
{
MorphoAttributePairs = EMPTY;
}
}
public IReadOnlyList <TermProbability> Run_Debug(string text)
{
if (text.IsNullOrWhiteSpace())
{
return(TERMPROBABILITY_EMPTY);
}
//---------------------------------------------------------//
var text_upper = StringsHelper.ToUpperInvariant(text); //---StringsHelper.ToUpperInvariantInPlace( text );
var tuples = _VTS.Run_Debug(text_upper);
return(tuples);
}
/// добавление слова и всех его форм в словарь
/// word - слово
/// pMorphoType - морфотип
/// nounType - тип сущетсвительного
unsafe public void AddWord(string word, MorphoType morphoType, MorphoAttributePair?nounType)
{
if (morphoType.MorphoForms.Length != 0)
{
var len = word.Length - morphoType.MorphoForms[0].Ending.Length;
var _base = (0 <= len) ? word.Substring(0, len) : word;
var baseMorphoForm = new BaseMorphoForm(_base, morphoType, nounType);
var _baseUpper = StringsHelper.ToUpperInvariant(_base);
fixed(char *baseUpper_ptr = _baseUpper)
{
AddWordPart(baseUpper_ptr, baseMorphoForm);
}
}
}
internal MorphoForm(string ending, List <MorphoAttributePair> morphoAttributePair)
{
_Ending = string.Intern(ending); //ending; //
_EndingUpper = string.Intern(StringsHelper.ToUpperInvariant(_Ending)); //StringsHelper.ToUpperInvariant( _Ending ); //
if (morphoAttributePair.Count != 0)
{
_MorphoAttributePairs = morphoAttributePair.ToArray();
}
else
{
_MorphoAttributePairs = EMPTY;
}
//_GlobalCount++;
//_GlobalHashsetEnding.Add( _Ending );
//_GlobalHashsetEndingUpper.Add( _EndingUpper );
}
/// добавление слова и всех его форм в словарь
/// word - слово
/// pMorphoType - морфотип
/// nounType - тип сущетсвительного
unsafe public void AddWord(string word, MorphoType morphoType, MorphoAttributePair?nounType)
{
#region
//if ( word == "он" )
//{
// System.Diagnostics.Debugger.Break();
//}
#endregion
if (morphoType.MorphoForms.Length != 0)
{
var len = word.Length - morphoType.MorphoForms[0].Ending.Length;
var _base = (0 <= len) ? word.Substring(0, len) : word;
var baseMorphoForm = new BaseMorphoForm(_base, morphoType, nounType);
var _baseUpper = StringsHelper.ToUpperInvariant(_base);
fixed(char *baseUpper_ptr = _baseUpper)
{
AddWordPart(baseUpper_ptr, baseMorphoForm);
}
}
}
public WordForms GetWordFormsByPartOfSpeech(string word, PartOfSpeechEnum partOfSpeechFilter)
{
var result = new WordForms(word);
var wordUpper = StringsHelper.ToUpperInvariant(word);
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);
}
unsafe public IReadOnlyList <TermProbability_Offset> Run_Offset(string text)
{
var textAsUpper = StringsHelper.ToUpperInvariant(text, out var isNullOrWhiteSpace);
if (isNullOrWhiteSpace)
{
return(TERMPROBABILITY_OFFSET_EMPTY);
}
//---------------------------------------------------------//
fixed(char *textAsUpper_ptr = textAsUpper)
{
var tuples = _VTS_Offset.Run(textAsUpper_ptr, textAsUpper.Length);
return(tuples);
}
#region comm. in-place to-upper
/*
* fixed ( char* text_ptr = text )
* {
* StringsHelper.ToUpperInvariantInPlace( text_ptr, out var isNullOrWhiteSpace );
* if ( isNullOrWhiteSpace )
* {
* return (TERMPROBABILITY_OFFSET_EMPTY);
* }
* //---------------------------------------------------------//
*
* var tuples = _VTS_Offset.Run( text_ptr, text.Length );
* return (tuples);
* }
*/
#endregion
}
public WordMorphology GetWordMorphology(string word, WordFormMorphologyModeEnum wordFormMorphologyMode)
{
var wordUpper = StringsHelper.ToUpperInvariant(word);
return(GetWordMorphology_NoToUpper(wordUpper, wordFormMorphologyMode));
}
/// <summary>
/// создание морфоформы из строки
/// </summary>
private MorphoFormNative?CreateMorphoForm(MorphoTypeNative morphoType, char *lineBase)
{
#region find index-of-COLON & check on length
var index = IndexOf(lineBase, COLON);
if ((index == -1) || (ENDING_BUFFER_SIZE <= index))
{
_ModelLoadingErrorCallback("Index of COLON is undefined or length the line is too long", StringsHelper.ToString(lineBase));
return(null);
}
#endregion
#region fill '_ENDING_LOWER_BUFFER'
var i = 0;
for (char *ptr = lineBase; i < index; ptr++, i++)
{
var ch = *ptr;
if ((_CHARTYPE_MAP[ch] & CharType.IsWhiteSpace) == CharType.IsWhiteSpace)
{
break;
}
_ENDING_LOWER_BUFFER[i] = _LOWER_INVARIANT_MAP[ch];
}
_ENDING_LOWER_BUFFER[i] = '\0';
#endregion
#region fill '_MorphoAttributePairs_Buffer'
_MorphoAttributePairs_Buffer.Clear();
for (char *ptr = lineBase + index + 1; ; ptr++)
{
var ch = *ptr;
if (ch == '\0')
{
break;
}
if ((_CHARTYPE_MAP[ch] & CharType.IsLetter) != CharType.IsLetter)
{
continue;
}
var len = 0;
for (; ; ptr++)
{
ch = *ptr;
if (ch == '\0')
{
break;
}
var ct = _CHARTYPE_MAP[ch];
if ((ct & CharType.IsLetter) != CharType.IsLetter &&
(ct & CharType.IsDigit) != CharType.IsDigit)
{
break;
}
len++;
}
if (len != 0)
{
var morphoAttribute = default(MorphoAttributeEnum);
if (_EnumParserMorphoAttribute.TryParse(ptr - len, len, ref morphoAttribute))
{
var map = _MorphoAttributeList.TryGetMorphoAttributePair(morphoType.MorphoAttributeGroup, morphoAttribute);
if (map.HasValue)
{
_MorphoAttributePairs_Buffer.Add(map.Value);
}
#if DEBUG
//*
//TOO MANY ERRORS AFTER last (2016.12.28) getting morpho-dcitionaries from 'LangAnalyzerStd-[ilook]'
else
{
var attr = new string( ptr - len, 0, len );
_ModelLoadingErrorCallback("Error in morpho-attribute: '" + attr + '\'', StringsHelper.ToString(lineBase));
}
//*/
#endif
}
else
{
var attr = new string(ptr - len, 0, len);
_ModelLoadingErrorCallback("Unknown morpho-attribute: '" + attr + '\'', StringsHelper.ToString(lineBase));
}
}
if (ch == '\0')
{
break;
}
}
#endregion
#region Allocate native-memory for endingOfWord
//*
IntPtr endingPtr;
IntPtr endingUpperPtr;
if ((i == 1) && (_ENDING_LOWER_BUFFER[0] == UNDERLINE))
{
endingPtr = _EMPTY_STRING;
endingUpperPtr = _EMPTY_STRING;
}
else
{
#region ending-in-original-case
endingPtr = new IntPtr(_ENDING_LOWER_BUFFER);
if (_EndingDictionary.TryGetValue(endingPtr, out IntPtr existsPtr))
{
endingPtr = existsPtr;
}
else
{
AllocHGlobalAndCopy(_ENDING_LOWER_BUFFER, index, out endingPtr);
_EndingDictionary.Add(endingPtr, endingPtr);
}
#endregion
#region ending-in-upper-case
StringsHelper.ToUpperInvariant(_ENDING_LOWER_BUFFER, _ENDING_UPPER_BUFFER);
endingUpperPtr = new IntPtr(_ENDING_UPPER_BUFFER);
if (_EndingDictionary.TryGetValue(endingUpperPtr, out existsPtr))
{
endingUpperPtr = existsPtr;
}
else
{
AllocHGlobalAndCopy(_ENDING_UPPER_BUFFER, index, out endingUpperPtr);
_EndingDictionary.Add(endingUpperPtr, endingUpperPtr);
}
#endregion
}
#endregion
var morphoForm = new MorphoFormNative((char *)endingPtr, (char *)endingUpperPtr, _MorphoAttributePairs_Buffer);
return(morphoForm);
}
/// получение морфологической информации
/// words - слова
public WordMorphology_t GetWordMorphology(string word)
{
var wordUpper = StringsHelper.ToUpperInvariant(word);
return(GetWordMorphology_NoToUpper(wordUpper, WordFormMorphologyModeEnum.Default));
}
public WordMorphology_t GetWordMorphology(string word, WordFormMorphologyModeEnum wordFormMorphologyMode) => GetWordMorphology_NoToUpper(StringsHelper.ToUpperInvariant(word), wordFormMorphologyMode);
