public Tokenizer(TokenizerConfig config)
{
_SentSplitter = new SentSplitter(config.SentSplitterConfig);
_Words = new List <word_t>(DEFAULT_WORDSLIST_CAPACITY);
_ParticleThatExclusion = config.Model.ParticleThatExclusion;
_SentSplitterProcessSentCallback_Delegate = new SentSplitter.ProcessSentCallbackDelegate(SentSplitterProcessSentCallback);
_UIM = xlat_Unsafe.Inst._UPPER_INVARIANT_MAP;
_CTM = xlat_Unsafe.Inst._CHARTYPE_MAP;
_CCTM = UnsafeConst.GetInstanceByLanguage(config.LanguageType)._CRF_CHARTYPE_MAP; //UnsafeConst.Inst._CRF_CHARTYPE_MAP;
//--//
ReAllocWordToUpperBuffer(DEFAULT_WORDTOUPPERBUFFER);
if ((config.TokenizeMode & TokenizeMode.PosTagger) == TokenizeMode.PosTagger)
{
_PosTaggerInputTypeProcessor = config.PosTaggerInputTypeProcessorFactory.CreateInstance();
//_Make_PosTagger = true;
}
else
{
_PosTaggerInputTypeProcessor = Dummy_PosTaggerInputTypeProcessor.Instance;
}
if ((config.TokenizeMode & TokenizeMode.Ner) == TokenizeMode.Ner)
{
_NerInputTypeProcessor = config.NerInputTypeProcessorFactory.CreateInstance();
//_Make_Ner = true;
}
else
{
_NerInputTypeProcessor = Dummy_NerInputTypeProcessor.Instance;
}
}
public NerInputTypeProcessorFactory(LanguageTypeEnum languageType)
{
switch (languageType)
{
case LanguageTypeEnum.En:
_NerInputTypeProcessor = new NerInputTypeProcessor_En();
break;
case LanguageTypeEnum.Ru:
_NerInputTypeProcessor = new NerInputTypeProcessor_Ru();
break;
default:
throw (new ArgumentException(languageType.ToString()));
}
}
private Tokenizer(TokenizerConfig4NerModelBuilder config)
{
config.UrlDetectorConfig.UrlExtractMode = UrlDetector.UrlExtractModeEnum.Position;
_urlDetector = new UrlDetector(config.UrlDetectorConfig);
_buildModelSentence = Sentence.CreateEmpty();
_words = new List <Word>(DEFAULT_WORDSLIST_CAPACITY);
_buildModelWords = new List <Buildmodel_word_t>(DEFAULT_WORDSLIST_CAPACITY);
_particleThatExclusion = config.Model.ParticleThatExclusion;
_UIM = XlatUnsafe.Inst._UPPER_INVARIANT_MAP;
_CTM = XlatUnsafe.Inst._CHARTYPE_MAP;
_CCTM = UnsafeConst.GetInstanceByLanguage(config.LanguageType)._CRF_CHARTYPE_MAP;
ReAllocWordToUpperBuffer(DEFAULT_WORDTOUPPERBUFFER);
_posTaggerInputTypeProcessor = DummyPosTaggerInputTypeProcessor.Instance;
_nerInputTypeProcessor = config.NerInputTypeProcessorFactory.CreateInstance();
}
private Tokenizer(TokenizerConfig4NerModelBuilder config)
{
_UrlDetector = new UrlDetector(new UrlDetectorConfig()
{
Model = config.UrlDetectorConfig.Model, UrlExtractMode = UrlDetector.UrlExtractModeEnum.Position
});
_BuildModelSent = sent_t.CreateEmpty();
_Words = new List <word_t>(DEFAULT_WORDSLIST_CAPACITY);
_BuildModelWords = new List <buildmodel_word_t>(DEFAULT_WORDSLIST_CAPACITY);
_ParticleThatExclusion = config.Model.ParticleThatExclusion;
_UIM = xlat_Unsafe.Inst._UPPER_INVARIANT_MAP;
_CTM = xlat_Unsafe.Inst._CHARTYPE_MAP;
_CCTM = UnsafeConst.GetInstanceByLanguage(config.LanguageType)._CRF_CHARTYPE_MAP; //UnsafeConst.Inst._CRF_CHARTYPE_MAP;
//--//
ReAllocWordToUpperBuffer(DEFAULT_WORDTOUPPERBUFFER);
_PosTaggerInputTypeProcessor = Dummy_PosTaggerInputTypeProcessor.Instance;
_NerInputTypeProcessor = config.NerInputTypeProcessorFactory.CreateInstance();
}
