public void Test1SentenceDictionaryOnly()
{
Analyzer analyzer = Analyzer.NewAnonymous(createComponents: (fieldname, reader) =>
{
var loader = new ClasspathResourceLoader(GetType());
var opennlpFactory = new OpenNLPTokenizerFactory(new Dictionary <string, string> {
{ "tokenizerModel", tokenizerModelFile }, { "sentenceModel", sentenceModelFile }
});
opennlpFactory.Inform(loader);
var opennlp = opennlpFactory.Create(reader);
var opennlpPOSFilterFactory = new OpenNLPPOSFilterFactory(new Dictionary <string, string> {
{ "posTaggerModel", posTaggerModelFile }
});
opennlpPOSFilterFactory.Inform(loader);
var opennlpPOSFilter = opennlpPOSFilterFactory.Create(opennlp);
var opennlpLemmatizerFilterFactory = new OpenNLPLemmatizerFilterFactory(new Dictionary <string, string> {
{ "dictionary", lemmatizerDictFile }
});
opennlpLemmatizerFilterFactory.Inform(loader);
var opennlpLemmatizerFilter = opennlpLemmatizerFilterFactory.Create(opennlpPOSFilter);
return(new TokenStreamComponents(opennlp, opennlpLemmatizerFilter));
});
//CustomAnalyzer analyzer = CustomAnalyzer.builder(new ClasspathResourceLoader(getClass()))
//.withTokenizer("opennlp", "tokenizerModel", tokenizerModelFile, "sentenceModel", sentenceModelFile)
//.addTokenFilter("opennlpPOS", "posTaggerModel", "en-test-pos-maxent.bin")
//.addTokenFilter("opennlplemmatizer", "dictionary", "en-test-lemmas.dict")
//.build();
AssertAnalyzesTo(analyzer, SENTENCE, SENTENCE_dict_punc, null, null,
SENTENCE_posTags, null, null, true);
}
public void TestBasic()
{
var loader = new ClasspathResourceLoader(GetType());
Analyzer analyzer = Analyzer.NewAnonymous(createComponents: (fieldname, reader) => {
var tokenizerFactory = new OpenNLPTokenizerFactory(new Dictionary <string, string> {
{ "tokenizerModel", tokenizerModelFile }, { "sentenceModel", sentenceModelFile }
});
tokenizerFactory.Inform(loader);
var tokenizer = tokenizerFactory.Create(reader);
var filter1Factory = new OpenNLPPOSFilterFactory(new Dictionary <string, string> {
{ "posTaggerModel", posTaggerModelFile }
});
filter1Factory.Inform(loader);
var filter1 = filter1Factory.Create(tokenizer);
return(new TokenStreamComponents(tokenizer, filter1));
});
// CustomAnalyzer analyzer = CustomAnalyzer.builder(new ClasspathResourceLoader(GetType()))
//.withTokenizer("opennlp", "tokenizerModel", tokenizerModelFile, "sentenceModel", sentenceModelFile)
//.addTokenFilter("opennlpPOS", "posTaggerModel", posTaggerModelFile)
//.build();
AssertAnalyzesTo(analyzer, SENTENCES, SENTENCES_punc, SENTENCES_startOffsets, SENTENCES_endOffsets);
}
public void TestKeywordAttributeAwarenessDictionaryAndMaxEnt()
{
Analyzer analyzer = Analyzer.NewAnonymous(createComponents: (fieldname, reader) =>
{
var loader = new ClasspathResourceLoader(GetType());
var opennlpFactory = new OpenNLPTokenizerFactory(new Dictionary <string, string> {
{ "tokenizerModel", tokenizerModelFile }, { "sentenceModel", sentenceModelFile }
});
opennlpFactory.Inform(loader);
var opennlp = opennlpFactory.Create(reader);
var opennlpPOSFilterFactory = new OpenNLPPOSFilterFactory(new Dictionary <string, string> {
{ "posTaggerModel", posTaggerModelFile }
});
opennlpPOSFilterFactory.Inform(loader);
var opennlpPOSFilter = opennlpPOSFilterFactory.Create(opennlp);
var keywordRepeatFilterFactory = new KeywordRepeatFilterFactory(new Dictionary <string, string>());
var keywordRepeatFilter = keywordRepeatFilterFactory.Create(opennlpPOSFilter);
var opennlpLemmatizerFilterFactory = new OpenNLPLemmatizerFilterFactory(new Dictionary <string, string> {
{ "dictionary", lemmatizerDictFile }, { "lemmatizerModel", lemmatizerModelFile }
});
opennlpLemmatizerFilterFactory.Inform(loader);
var opennlpLemmatizerFilter = opennlpLemmatizerFilterFactory.Create(keywordRepeatFilter);
var removeDuplicatesTokenFilterFactory = new RemoveDuplicatesTokenFilterFactory(new Dictionary <string, string>());
var removeDuplicatesTokenFilter = removeDuplicatesTokenFilterFactory.Create(opennlpLemmatizerFilter);
return(new TokenStreamComponents(opennlp, removeDuplicatesTokenFilter));
});
//CustomAnalyzer analyzer = CustomAnalyzer.builder(new ClasspathResourceLoader(getClass()))
// .withTokenizer("opennlp", "tokenizerModel", tokenizerModelFile, "sentenceModel", sentenceModelFile)
// .addTokenFilter("opennlpPOS", "posTaggerModel", posTaggerModelFile)
// .addTokenFilter(KeywordRepeatFilterFactory.class)
// .addTokenFilter("opennlplemmatizer", "dictionary", lemmatizerDictFile, "lemmatizerModel", lemmatizerModelFile)
// .addTokenFilter(RemoveDuplicatesTokenFilterFactory.class)
// .build();
AssertAnalyzesTo(analyzer, SENTENCES_both, SENTENCES_both_keep_orig_punc, null, null,
SENTENCES_both_keep_orig_posTags, null, null, true);
}
public void TestPayloads()
{
//CustomAnalyzer analyzer = CustomAnalyzer.builder(new ClasspathResourceLoader(getClass()))
//.withTokenizer("opennlp", "tokenizerModel", tokenizerModelFile, "sentenceModel", sentenceModelFile)
//.addTokenFilter("opennlpPOS", "posTaggerModel", posTaggerModelFile)
//.addTokenFilter("opennlpChunker", "chunkerModel", chunkerModelFile)
//.addTokenFilter(TypeAsPayloadTokenFilterFactory.class)
//.build();
Analyzer analyzer = Analyzer.NewAnonymous(createComponents: (fieldName, reader) =>
{
var loader = new ClasspathResourceLoader(GetType());
var opennlpFactory = new OpenNLPTokenizerFactory(new Dictionary <string, string> {
{ "tokenizerModel", tokenizerModelFile }, { "sentenceModel", sentenceModelFile }
});
opennlpFactory.Inform(loader);
var opennlp = opennlpFactory.Create(NewAttributeFactory(), reader); //new OpenNLPTokenizer(reader, new Tools.NLPSentenceDetectorOp(sentenceModelFile), new Tools.NLPTokenizerOp(tokenizerModelFile));
var opennlpPOSFilterFactory = new OpenNLPPOSFilterFactory(new Dictionary <string, string> {
{ "posTaggerModel", posTaggerModelFile }
});
opennlpPOSFilterFactory.Inform(loader);
var opennlpPOSFilter = opennlpPOSFilterFactory.Create(opennlp); //new OpenNLPPOSFilter(opennlp, new Tools.NLPPOSTaggerOp(posTaggerModelFile));
var opennlpChunkerFilterFactory = new OpenNLPChunkerFilterFactory(new Dictionary <string, string> {
{ "chunkerModel", chunkerModelFile }
});
opennlpChunkerFilterFactory.Inform(loader);
var opennlpChunkerFilter = opennlpChunkerFilterFactory.Create(opennlpPOSFilter); //new OpenNLPChunkerFilter(filter1, new Tools.NLPChunkerOp(chunkerModelFile));
var typeAsPayloadFilterFactory = new TypeAsPayloadTokenFilterFactory(new Dictionary <string, string>());
var typeAsPayloadFilter = typeAsPayloadFilterFactory.Create(opennlpChunkerFilter);
return(new TokenStreamComponents(opennlp, typeAsPayloadFilter));
});
AssertAnalyzesTo(analyzer, SENTENCES, SENTENCES_punc, SENTENCES_startOffsets, SENTENCES_endOffsets,
null, null, null, true, ToPayloads(SENTENCES_chunks));
}
