public void Dispose()
{
if (Processor != null)
{
Processor.Dispose();
Processor = null;
}
if (MorphoModel != null)
{
MorphoModel.Dispose();
MorphoModel = null;
}
if (MorphoAmbiguityResolverModel != null)
{
MorphoAmbiguityResolverModel.Dispose();
MorphoAmbiguityResolverModel = null;
}
if (SentSplitterConfig != null)
{
SentSplitterConfig.Dispose();
SentSplitterConfig = null;
}
}
public static ConcurrentFactory GetConcurrentFactory()
{
var f = _ConcurrentFactory;
if (f == null)
{
lock ( _SyncLock )
{
f = _ConcurrentFactory;
if (f == null)
{
var sentSplitterConfig = new SentSplitterConfig(Config.SENT_SPLITTER_RESOURCES_XML_FILENAME,
Config.URL_DETECTOR_RESOURCES_XML_FILENAME);
var config = new NerProcessorConfig(Config.TOKENIZER_RESOURCES_XML_FILENAME,
Config.LANGUAGE_TYPE,
sentSplitterConfig)
{
ModelFilename = Config.NER_MODEL_FILENAME,
TemplateFilename = Config.NER_TEMPLATE_FILENAME,
};
f = new ConcurrentFactory(config, Config.CONCURRENT_FACTORY_INSTANCE_COUNT);
_ConcurrentFactory = f;
}
}
}
return(f);
}
public static (PosTaggerProcessorConfig config, SentSplitterConfig sentSplitterConfig) CreatePosTaggerProcessorConfig()
{
var sentSplitterConfig = new SentSplitterConfig(Config.SENT_SPLITTER_RESOURCES_XML_FILENAME, Config.URL_DETECTOR_RESOURCES_XML_FILENAME);
var config = new PosTaggerProcessorConfig(Config.TOKENIZER_RESOURCES_XML_FILENAME, Config.POSTAGGER_RESOURCES_XML_FILENAME, LanguageTypeEnum.Ru, sentSplitterConfig)
{
ModelFilename = Config.POSTAGGER_MODEL_FILENAME,
TemplateFilename = Config.POSTAGGER_TEMPLATE_FILENAME,
};
return(config, sentSplitterConfig);
}
public NerProcessorConfig(string tokenizerResourcesXmlFilename,
LanguageTypeEnum languageType,
SentSplitterConfig sentSplitterConfig)
{
TokenizerConfig = new TokenizerConfig(tokenizerResourcesXmlFilename)
{
TokenizeMode = TokenizeMode.Ner,
SentSplitterConfig = sentSplitterConfig,
NerInputTypeProcessorFactory = new NerInputTypeProcessorFactory(languageType),
LanguageType = languageType,
};
}
private static NerProcessorConfig CreateNerProcessorConfig()
{
var sentSplitterConfig = new SentSplitterConfig(Config.SENT_SPLITTER_RESOURCES_XML_FILENAME,
Config.URL_DETECTOR_RESOURCES_XML_FILENAME);
var config = new NerProcessorConfig(Config.TOKENIZER_RESOURCES_XML_FILENAME,
Config.LANGUAGE_TYPE, sentSplitterConfig)
{
ModelFilename = Config.NER_MODEL_FILENAME,
TemplateFilename = Config.NER_TEMPLATE_FILENAME,
};
return(config);
}
public PosTaggerProcessorConfig(string tokenizerResourcesXmlFilename,
string posTaggerResourcesXmlFilename,
LanguageTypeEnum languageType,
SentSplitterConfig sentSplitterConfig)
{
Model = new PosTaggerResourcesModel(posTaggerResourcesXmlFilename);
TokenizerConfig = new TokenizerConfig(tokenizerResourcesXmlFilename)
{
TokenizeMode = TokenizeMode.PosTagger,
LanguageType = languageType,
SentSplitterConfig = sentSplitterConfig,
PosTaggerInputTypeProcessorFactory = new PosTaggerInputTypeProcessorFactory(Model, languageType),
};
}
public ner_tokenizer(SentSplitterConfig config, int wordCapacity)
{
_SentSplitter = new SentSplitter(config);
_Words = new List <word_t>(wordCapacity);
}
private bool _NotSkipNonLetterAndNonDigitToTheEnd; //need for NER-model-builder
#endregion
public ner_tokenizer(SentSplitterConfig config)
{
_SentSplitter = new SentSplitter(config);
_Words = new List <word_t>(DEFAULT_WORDSLIST_CAPACITY);
}
