public SentSplitterConfig(string sentSplitterResourcesXmlFilename,
string urlDetectorResourcesXmlFilename)
{
Model = new SentSplitterModel(sentSplitterResourcesXmlFilename);
UrlDetectorConfig = new UrlDetectorConfig(urlDetectorResourcesXmlFilename);
SplitBySmiles = true;
}
public ner_tokenizer(UrlDetectorConfig config)
{
config.UrlExtractMode = UrlDetector.UrlExtractModeEnum.Position;
_UrlDetector = new UrlDetector(config);
_BuildModelSent = sent_t.CreateEmpty();
_Words = new List <word_t>(DEFAULT_WORDSLIST_CAPACITY);
_BuildModelWords = new List <buildmodel_word_t>(DEFAULT_WORDSLIST_CAPACITY);
}
public NerModelBuilderConfig(string tokenizerResourcesXmlFilename,
LanguageTypeEnum languageType,
UrlDetectorConfig urlDetectorConfig)
{
TokenizerConfig4NerModelBuilder = new TokenizerConfig4NerModelBuilder(tokenizerResourcesXmlFilename)
{
UrlDetectorConfig = urlDetectorConfig,
NerInputTypeProcessorFactory = new NerInputTypeProcessorFactory(languageType),
LanguageType = languageType,
};
}
public PosTaggerModelBuilder(string templateFilename,
LanguageTypeEnum languageType,
UrlDetectorConfig urlDetectorConfig)
{
templateFilename.ThrowIfNullOrWhiteSpace("templateFilename");
urlDetectorConfig.ThrowIfNull("urlDetectorConfig");
_posTaggerScriber = PosTaggerScriber.Create4ModelBuilder(templateFilename);
_posTaggerInputTypeProcessor = CreatePosTaggerInputTypeProcessor(languageType);
_urlDetector = new UrlDetector(urlDetectorConfig);
_words = new List <Word>();
}
public ConcurrentFactory(UrlDetectorConfig config, int instanceCount)
{
if (instanceCount <= 0)
{
throw (new ArgumentException("instanceCount"));
}
_Semaphore = new Semaphore(instanceCount, instanceCount);
_Stack = new ConcurrentStack <UrlDetector>();
for (int i = 0; i < instanceCount; i++)
{
_Stack.Push(new UrlDetector(config));
}
}
public mld_tokenizer(UrlDetectorModel urlModel, int wordCapacity)
{
var urlConfig = new UrlDetectorConfig()
{
Model = urlModel,
UrlExtractMode = UrlDetector.UrlExtractModeEnum.Position,
};
_UrlDetector = new UrlDetector(urlConfig);
_Words = new List <string>(Math.Max(DEFAULT_WORDCAPACITY, wordCapacity));
_NgramsSB = new StringBuilder();
_AddWordToListAction = new Action <string>(AddWordToList);
_UIM = xlat_Unsafe.Inst._UPPER_INVARIANT_MAP;
_CTM = xlat_Unsafe.Inst._CHARTYPE_MAP;
_IAW = UnsafeConst.Inst._INTERPRETE_AS_WHITESPACE;
_DWC = UnsafeConst.Inst._DIGIT_WORD_CHARS;
//--//
ReAllocWordToUpperBuffer(DEFAULT_WORDTOUPPERBUFFER);
}
public ConcurrentFactory GetConcurrentFactory()
{
var f = _ConcurrentFactory;
if (f == null)
{
lock ( _SyncLock )
{
f = _ConcurrentFactory;
if (f == null)
{
var config = new UrlDetectorConfig(Config.URL_DETECTOR_RESOURCES_XML_FILENAME)
{
UrlExtractMode = UrlDetector.UrlExtractModeEnum.Position,
};
f = new ConcurrentFactory(config, Config.CONCURRENT_FACTORY_INSTANCE_COUNT);
_ConcurrentFactory = f;
}
}
}
return(f);
}