org.deeplearning4j.text.tokenization.tokenizer.PosUimaTokenizer Maven / Gradle / Ivy
/*******************************************************************************
* Copyright (c) 2015-2018 Skymind, Inc.
*
* This program and the accompanying materials are made available under the
* terms of the Apache License, Version 2.0 which is available at
* https://www.apache.org/licenses/LICENSE-2.0.
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS, WITHOUT
* WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the
* License for the specific language governing permissions and limitations
* under the License.
*
* SPDX-License-Identifier: Apache-2.0
******************************************************************************/
package org.deeplearning4j.text.tokenization.tokenizer;
import lombok.NonNull;
import org.apache.uima.analysis_engine.AnalysisEngine;
import org.apache.uima.cas.CAS;
import org.apache.uima.fit.factory.AnalysisEngineFactory;
import org.apache.uima.fit.util.JCasUtil;
import org.cleartk.token.type.Sentence;
import org.cleartk.token.type.Token;
import org.deeplearning4j.text.annotator.PoStagger;
import org.deeplearning4j.text.annotator.SentenceAnnotator;
import org.deeplearning4j.text.annotator.StemmerAnnotator;
import org.deeplearning4j.text.annotator.TokenizerAnnotator;
import java.util.ArrayList;
import java.util.Collection;
import java.util.List;
/**
* Filter by part of speech tag.
* Any not valid part of speech tags
* become NONE
* @author Adam Gibson
*
*/
public class PosUimaTokenizer implements Tokenizer {
private static AnalysisEngine engine;
private List tokens;
private Collection allowedPosTags;
private int index;
private static CAS cas;
private TokenPreProcess preProcessor;
private boolean stripNones = false;
public PosUimaTokenizer(String tokens, AnalysisEngine engine, Collection allowedPosTags) {
this(tokens, engine, allowedPosTags, false);
}
public PosUimaTokenizer(String tokens, AnalysisEngine engine, Collection allowedPosTags,
boolean stripNones) {
if (PosUimaTokenizer.engine == null)
PosUimaTokenizer.engine = engine;
this.allowedPosTags = allowedPosTags;
this.tokens = new ArrayList<>();
this.stripNones = stripNones;
try {
if (cas == null)
cas = engine.newCAS();
cas.reset();
cas.setDocumentText(tokens);
PosUimaTokenizer.engine.process(cas);
for (Sentence s : JCasUtil.select(cas.getJCas(), Sentence.class)) {
for (Token t : JCasUtil.selectCovered(Token.class, s)) {
//add NONE for each invalid token
if (valid(t))
if (t.getLemma() != null)
this.tokens.add(t.getLemma());
else if (t.getStem() != null)
this.tokens.add(t.getStem());
else
this.tokens.add(t.getCoveredText());
else
this.tokens.add("NONE");
}
}
} catch (Exception e) {
throw new RuntimeException(e);
}
}
private boolean valid(Token token) {
String check = token.getCoveredText();
if (check.matches("<[A-Z]+>") || check.matches("")
|| (token.getPos() != null && !this.allowedPosTags.contains(token.getPos())))
return false;
return true;
}
@Override
public boolean hasMoreTokens() {
return index < tokens.size();
}
@Override
public int countTokens() {
return tokens.size();
}
@Override
public String nextToken() {
String ret = tokens.get(index); // preProcessor != null ? preProcessor.preProcess(tokens.get(index)) : tokens.get(index);
index++;
return ret;
}
@Override
public List getTokens() {
List tokens = new ArrayList<>();
while (hasMoreTokens()) {
String nextT = nextToken();
if (stripNones && nextT.equals("NONE"))
continue;
tokens.add(preProcessor != null ? preProcessor.preProcess(nextT) : nextT);
}
return tokens;
}
public static AnalysisEngine defaultAnalysisEngine() {
try {
return AnalysisEngineFactory.createEngine(AnalysisEngineFactory.createEngineDescription(
SentenceAnnotator.getDescription(), TokenizerAnnotator.getDescription(),
PoStagger.getDescription("en"), StemmerAnnotator.getDescription("English")));
} catch (Exception e) {
throw new RuntimeException(e);
}
}
@Override
public void setTokenPreProcessor(@NonNull TokenPreProcess tokenPreProcessor) {
this.preProcessor = tokenPreProcessor;
}
}