Java JCasUtil.select方法代码示例

import org.uimafit.util.JCasUtil; //导入方法依赖的package包/类
protected Map<String, String> indexDepTree(JCas text) {
	Map<String, String> depTree = new HashMap<String, String>();

	// format: key: 1 ### word ### pos; value: dep_rel ## 2 ### word ### pos
	// escape: .replace("#", "\\#")
	// depTree.put("1 ### The ### Det", "DET ## 2 ### dog ### N");
	// depTree.put("2 ### dog ### N", "SUBJ ## 3 ### chases ### V");
	// depTree.put("3 ### chases ### V", "ROOT ## 0 ### NULL ### NULL");
	// depTree.put("4 ### The ### Det", "DET ## 5 ### cat ### N");
	// depTree.put("5 ### cat ### N", "OBJ ## 3 ### chases ### V");
	for (Dependency dep : JCasUtil.select(text, Dependency.class)) {
		Token child = dep.getDependent();
		Token parent = dep.getGovernor();
		depTree.put(child.getBegin() + " ### "
				+ child.getCoveredText().replace("#", "\\#") + " ### "
				+ child.getPos().getPosValue(), dep.getDependencyType()
				+ " ## " + parent.getBegin() + " ### "
				+ parent.getCoveredText().replace("#", "\\#") + " ### "
				+ parent.getPos().getPosValue());
	}

	return depTree;
}
 

import org.uimafit.util.JCasUtil; //导入方法依赖的package包/类
protected Map<String, String> indexLemmaDepTree(JCas text) {
	Map<String, String> depTree = new HashMap<String, String>();

	for (Dependency dep : JCasUtil.select(text, Dependency.class)) {
		Token child = dep.getDependent();
		Token parent = dep.getGovernor();
		depTree.put(child.getBegin() + " ### "
				+ child.getLemma().getValue().replace("#", "\\#") + " ### "
				+ child.getPos().getPosValue(), dep.getDependencyType()
				+ " ## " + parent.getBegin() + " ### "
				+ parent.getLemma().getValue().replace("#", "\\#")
				+ " ### " + parent.getPos().getPosValue());
	}

	return depTree;
}
 

import org.uimafit.util.JCasUtil; //导入方法依赖的package包/类
/**
 * 
 * @param jcas a JCas created by an EOP LAP
 * @return A list of roots, each belonging to a BIU dependency parse tree of a sentence in the CAS. Trees are ordered by sentence order.
 * @throws CasTreeConverterException
 * @throws UnsupportedPosTagStringException
 */
public List<BasicNode> convertCasToTrees(JCas jcas) throws CasTreeConverterException, UnsupportedPosTagStringException {
	Collection<Sentence> sentenceAnnotations = JCasUtil.select(jcas, Sentence.class);
	lastSentenceList = new ArrayList<Sentence>(sentenceAnnotations);
	lastTokenToNodeBySentence = new LinkedHashMap<Sentence, OneToManyBidiMultiHashMap<Token,BasicNode>>();
	lastRootList = new ArrayList<BasicNode>(sentenceAnnotations.size());
	lastSentenceToRoot = new DualHashBidiMap<>();
	
	for (Sentence sentenceAnno : lastSentenceList) {
		BasicNode root = convertSentenceToTree(jcas, sentenceAnno);
		lastSentenceToRoot.put(sentenceAnno, root);
		lastRootList.add(root);
		
		OneToManyBidiMultiHashMap<Token,BasicNode> tokenToNodeCopy = new OneToManyBidiMultiHashMap<Token,BasicNode>(tokenToNode);
		lastTokenToNodeBySentence.put(sentenceAnno, tokenToNodeCopy);
	}
	
	return lastRootList;
}
 

import org.uimafit.util.JCasUtil; //导入方法依赖的package包/类
@Override
public void process(JCas aCAS) throws AnalysisEngineProcessException {
	Token t;
	this.jcas=aCAS;
	for(Substitution s : JCasUtil.select(jcas, Substitution.class)) {
		System.out.println(s.toString());
		assert(s.getSenseclass().equals("[email protected]@1"));
		assert(s.getBegin()==37);
	}

}
 

import org.uimafit.util.JCasUtil; //导入方法依赖的package包/类
/**
 * This utility method fetches alignment.Link instances that links the give 
 * "Type" of annotations. More specifically, the method returns all link 
 * instances that connects Targets, which holds the give "type". 
 * 
 * For example, a call with type=Token.class will return all Link instances 
 * where either of its TSideTarget or HSideTarget holds "Token" annotation. 
 * 
 * The method will return all link instances, if one of its Targets hold 
 * the given the type.  
 * 
 * @param jCas the JCas with EOP views 
 * @param type target annotation class. 
 * @return a List<Link> that holds all links that satisfy the condition. If none satisfy the condition, it will return an empty List. 
 */
public static <T extends TOP> List<Link> selectLinksWith(JCas aJCas, Class<T> type) throws CASException 
{
	List<Link> resultList = new ArrayList<Link>(); 
	
	JCas hypoView = aJCas.getView("HypothesisView");
	// get Links that satisfy the condition by iterating all Links just once. 
	
	for (Link l : JCasUtil.select(hypoView, Link.class)) 
	{
		// is this link holds type object in either of its target? 
		Target tt = l.getTSideTarget(); 
		Target ht = l.getHSideTarget(); 
		
		if (JCasUtil.select(tt.getTargetAnnotations(), type).size() > 0)
		{
			// T side target does hold at least one of type instance. 
			resultList.add(l); 
			continue; // no need to check h side target 
		}
		
		if (JCasUtil.select(ht.getTargetAnnotations(), type).size() > 0)
		{
			// H side target does hold at least one of type instance. 
			resultList.add(l); 
		}
	}
	return resultList; 
}
 

import org.uimafit.util.JCasUtil; //导入方法依赖的package包/类
/**
 * This method is a helper utility that is required to look up Meteor Phrase tables. 
 * 
 * Basically, returns all possible phrase candidates up to N words in a List<String> 
 * 
 * The method uses Token annotation in JCas to generate possible candidates. Thus, 
 * a tokenization annotator should have annotated this JCas before.  
 * 
 * @param JCas aJCas The view, that holds the sentence(s) to be analyzed. 
 * @param int uptoN The maximum number of 
 * @return 
 */
public static List<String> getPhraseCandidatesFromSOFA(JCas aJCas, int uptoN)
{
	// sanity check 
	assert(aJCas !=null); 
	assert(uptoN > 0); 
	
	// list for result, 
	List<String> result = new ArrayList<String>(); 
	
	// ok. work. first; 
	// get all Tokens (by appearing orders...) 
	Collection<Token> t = JCasUtil.select(aJCas, Token.class); 
	Token[] tokens = t.toArray(new Token[t.size()]); 
	
	// then; 
	// for each Token, start uptoN process. 
	for(int i=0; i < tokens.length; i++)
	{
		for(int j=0; (j < uptoN) && (i+j < tokens.length); j++ )
		{
			Token leftEnd = tokens[i]; 
			Token rightEnd = tokens[i+j]; 
			String text = aJCas.getDocumentText().substring(leftEnd.getBegin(), rightEnd.getEnd()); 
			// and store in lower case. 
			result.add(text.toLowerCase()); 
		}			
	}
	
	// done 
	// all candidates are store here. 
	return result; 
}
 

import org.uimafit.util.JCasUtil; //导入方法依赖的package包/类
private static List<Token> tokensBetween(JCas aJCas, int from, int to)
{
	List<Token> tokenList = new ArrayList<Token>(); 
	
	for (Token token: JCasUtil.select(aJCas, Token.class))
	{
		if ( (token.getBegin() >= from) && (token.getEnd() <= to))
		{
			tokenList.add(token); 
		}
	}
	return tokenList; 
}
 

import org.uimafit.util.JCasUtil; //导入方法依赖的package包/类
/**
 * Uses the annotations in the CAS and extracts the tokens and 
 * their lemmas from the text and hypothesis views
 * @param aJCas The JCas object of the text and hypothesis, 
 * after tokenization and lemmatization.
 * @throws CASException
 */
private void getTokenAnnotations(JCas aJCas) throws CASException {
	
	// Get the text and hypothesis views
	JCas textView = aJCas.getView(LAP_ImplBase.TEXTVIEW);
	JCas hypoView = aJCas.getView(LAP_ImplBase.HYPOTHESISVIEW);
		
	// Get the tokens
	textTokens = new ArrayList<Token>(JCasUtil.select(textView, Token.class));
	hypoTokens = new ArrayList<Token>(JCasUtil.select(hypoView, Token.class));
}
 

import org.uimafit.util.JCasUtil; //导入方法依赖的package包/类
@Override
public void process(JCas jcas) throws AnalysisEngineProcessException {
	try {
		CAS cas = jcas.getCas();
		mappingProvider.configure(cas);
		
		for (Sentence sentenceAnno : JCasUtil.select(jcas, Sentence.class)) {
			List<Token> tokens = JCasUtil.selectCovered(jcas, Token.class, sentenceAnno);
			List<String> tokenStrings = JCasUtil.toText(tokens);
			List<PosTaggedToken> taggedTokens;
			
			synchronized (innerTool) {
				innerTool.setTokenizedSentence(tokenStrings);
				innerTool.process();
				taggedTokens = innerTool.getPosTaggedTokens();
			}
			
			if (taggedTokens.size() != tokens.size()) {
				throw new PosTaggerException("Got pos tagging for " + taggedTokens.size() +
						" tokens, should have gotten according to the total number of tokens in the sentence: " + tokens.size());
			}
			
			Iterator<Token> tokenIter = tokens.iterator();
			for (PosTaggedToken taggedToken : taggedTokens) {
				Token tokenAnno = tokenIter.next();
				String tagString = taggedToken.getPartOfSpeech().getStringRepresentation();
				
				// Get an annotation with the appropriate UIMA type via the mappingProvider
				Type posTag = mappingProvider.getTagType(tagString);
				POS posAnnotation = (POS) cas.createAnnotation(posTag, tokenAnno.getBegin(), tokenAnno.getEnd());
				posAnnotation.setPosValue(tagString);
				posAnnotation.addToIndexes();
				
				tokenAnno.setPos(posAnnotation);
			}
		}
	} catch (PosTaggerException e) {
		throw new AnalysisEngineProcessException(AnalysisEngineProcessException.ANNOTATOR_EXCEPTION, null, e);
	}
}
 

import org.uimafit.util.JCasUtil; //导入方法依赖的package包/类
private static void assertTokenToNodes(JCas jcas, List<BasicNode> testedTrees, Map<Sentence, OneToManyBidiMultiHashMap<Token, BasicNode>> tokenToNodesBySentence) {
	Collection<Sentence> sentences = JCasUtil.select(jcas, Sentence.class);
	assertOrderedFeatureStructureCollectionsEqual(sentences, tokenToNodesBySentence.keySet());
	
	assertEquals(testedTrees.size(), tokenToNodesBySentence.size());
	
	Iterator<BasicNode> rootIter = testedTrees.iterator();
	for (Entry<Sentence, OneToManyBidiMultiHashMap<Token, BasicNode>> entry : tokenToNodesBySentence.entrySet()) {
		OneToManyBidiMultiHashMap<Token, BasicNode> tokenToNodes = entry.getValue();
		
		List<Token> tokens = JCasUtil.selectCovered(jcas, Token.class, entry.getKey());
		assertUnorderedFeatureStructureCollectionsEqual(tokens, tokenToNodes.keySet());
		
		BasicNode root = rootIter.next();
		Set<BasicNode> nodes = AbstractNodeUtils.treeToSet(root);
		
		// Remove null-word nodes, since this is the "fake" root node, and we don't need it
		// for the comparison (it is not present in the tokenToNodes map)
		for (Iterator<BasicNode> iterNodes = nodes.iterator(); iterNodes.hasNext();) {
			if (iterNodes.next().getInfo().getNodeInfo().getWord()==null) {
				iterNodes.remove();
			}
		}
		assertEquals(nodes, new HashSet<BasicNode>(tokenToNodes.values()));
		
		for (Entry<Token, Collection<BasicNode>> tokenNodesEntry : tokenToNodes.entrySet()) {
			List<BasicNode> nodeList = (List<BasicNode>) tokenNodesEntry.getValue();
			for (int i=0; i<nodeList.size(); i++) {
				assertEquals(tokenNodesEntry.getKey().getLemma().getValue(), nodeList.get(i).getInfo().getNodeInfo().getWordLemma());
				if (i==0) {
					assertNull(nodeList.get(i).getAntecedent()); // only first element is non-deep node (no antecedent)
				}
				else {
					assertNotNull(nodeList.get(i).getAntecedent()); // all other element are deep (have antecedent)
				}
			}
		}
	}
}
 

import org.uimafit.util.JCasUtil; //导入方法依赖的package包/类
private Map<String,Link> getAlignments(JCas jcas) throws Exception {
	
   	Map<String,Link> result = new HashMap<String,Link>();
   	
	try {
		
		// Call the aligner component to get the alignments between T and H
		if (this.aligner != null) {
			logger.finer("\ngetting the alignments ...");
			aligner.annotate(jcas);
			logger.finer("done.");
		}
					
		logger.finer("\n\nalignments list:\n");
		
		//get the HYPOTHESIS view
		JCas hypoView = jcas.getView(LAP_ImplBase.HYPOTHESISVIEW);
		
		//cycle through the alignments
		for (Link link : JCasUtil.select(hypoView, Link.class)) {
			
			logger.finer(String.format("\nText phrase: %s, " +
					"Hypothesis phrase: %s, " + 
					"id: %s, confidence: %f, direction: %s", 
					link.getTSideTarget().getCoveredText(),
					link.getHSideTarget().getCoveredText(),
					link.getID(), link.getStrength(),
					link.getDirection().toString())); 
			
			String key = link.getTSideTarget().getCoveredText() +
			        		"__" + 
			        		link.getHSideTarget().getCoveredText();
			     
			//for a couple of tokens it can save a type of alignment only in the
			//order as provided by the aligner component.
			if (!result.containsKey(key))
				result.put(key, link);
			
		}
		
	} catch (Exception e) {
		
		throw new Exception(e.getMessage());
		
	}
	
	return result;
}
 

import org.uimafit.util.JCasUtil; //导入方法依赖的package包/类
@Test
public void test1() {
	
	try {
		
		// Annotate the first pair with tokens and lemmas
     	logger.info("Tokenize and lemmatize the sentence pair #1");
		JCas pair1 = lap.generateSingleTHPairCAS(t1, h1);
		
		// Call the aligner to align T and H of pair 1
		logger.info("Aligning pair #1");
		aligner.annotate(pair1);
		logger.info("Finished aligning pair #1");
					
		// Print the alignment of pair 1
		JCas hypoView = pair1.getView(LAP_ImplBase.HYPOTHESISVIEW);
		
		logger.info("Pair 1:");
		logger.info("T: " + t1); 
		logger.info("H: " + h1);
		
		boolean assassinKiller = false, deathPenaltyCapitalPunishment = false;
		
		for (Link link : JCasUtil.select(hypoView, Link.class)) {
			
			logger.info(String.format("Text phrase: %s, " +
					"hypothesis phrase: %s, " + 
					"id: %s, confidence: %f, direction: %s", 
					link.getTSideTarget().getCoveredText(),
					link.getHSideTarget().getCoveredText(),
					link.getID(), link.getStrength(),
					link.getDirection().toString()));
			
			assassinKiller = assassinKiller || ((link.getTSideTarget().getBegin() == 4) &&
												(link.getTSideTarget().getEnd() == 12) && 
												(link.getHSideTarget().getBegin() == 4) &&
												(link.getHSideTarget().getEnd() == 10));
			
			deathPenaltyCapitalPunishment = deathPenaltyCapitalPunishment || 
					((link.getTSideTarget().getBegin() == 44) &&
					(link.getTSideTarget().getEnd() == 57) && 
					(link.getHSideTarget().getBegin() == 52) &&
					(link.getHSideTarget().getEnd() == 70));
		}
		
		// Make sure the alignments contain the alignment of
		// "assassin" to "killer"
		if (!assassinKiller) {
			fail("There is no alignment link between 'assassin' and 'killer'." +
					" This alignment link appears in: WordNet, BAP, Lin Dependency," +
					" Lin Proximity and Wikipedia. " +
					" Make sure that at least some of these resources were loaded correctly.");
		}
		
		// Make sure the alignments contain the alignment of
		// "death penalty" to "capital punishment"
		if (!deathPenaltyCapitalPunishment) {
			fail("There is no alignment link between 'death penalty' and 'capital punishment'." +
					" This alignment link appears in: WordNet and Wikipedia. " +
					" Make sure that at least one of these resources was loaded correctly.");
		}
		
	} catch (Exception e) {
		logger.info("Could not process first pair. " + e.getMessage());
	}
}
 

import org.uimafit.util.JCasUtil; //导入方法依赖的package包/类
@Test
public void test2() {
	
	try {
		
		// Annotate the first pair with tokens and lemmas
     	logger.info("Tokenize and lemmatize the sentence pair #2");
     	JCas pair2 = lap.generateSingleTHPairCAS(t2, h2);
		
		// Call the aligner to align T and H of pair 2
		logger.info("Aligning pair #2");
		aligner.annotate(pair2);
		logger.info("Finished aligning pair #2");
					
		// Print the alignment of pair 1
		JCas hypoView = pair2.getView(LAP_ImplBase.HYPOTHESISVIEW);
		
		logger.info("Pair 2:");
		logger.info("T: " + t2); 
		logger.info("H: " + h2);
		
		boolean killedWounded = false, dallasTexas = false;
		
		for (Link link : JCasUtil.select(hypoView, Link.class)) {
			
			logger.info(String.format("Text phrase: %s, " +
								"hypothesis phrase: %s, " + 
								"id: %s, confidence: %f, direction: %s", 
								link.getTSideTarget().getCoveredText(),
								link.getHSideTarget().getCoveredText(),
								link.getID(), link.getStrength(),
								link.getDirection().toString()));
			
			killedWounded = killedWounded || ((link.getTSideTarget().getBegin() == 12) &&
					(link.getTSideTarget().getEnd() == 18) && 
					(link.getHSideTarget().getBegin() == 12) &&
					(link.getHSideTarget().getEnd() == 19));

			dallasTexas = dallasTexas || ((link.getTSideTarget().getBegin() == 22) &&
					(link.getTSideTarget().getEnd() == 28) && 
					(link.getHSideTarget().getBegin() == 32) &&
					(link.getHSideTarget().getEnd() == 37));
		}
		
		// Make sure the alignments contain the alignment of
		// "killed" to "wounded"
		if (!killedWounded) {
			fail("There is no alignment link between 'killed' and 'wounded'." +
					" This alignment link appears in VerbOcean." +
					" Make sure that this resource was loaded correctly.");
		}
		
		// Make sure the alignments contain the alignment of
		// "Dallas" to "Texas"
		if (!dallasTexas) {
			fail("There is no alignment link between 'Dallas' and 'Texas'." +
					" This alignment link appears in: WordNet and GEO. " +
					" Make sure that at least one of these resources was loaded correctly.");
		}
		
	} catch (Exception e) {
		logger.info("Could not process first pair. " + e.getMessage());
	}
}
 

import org.uimafit.util.JCasUtil; //导入方法依赖的package包/类
@Override
public void process(JCas jcas) throws AnalysisEngineProcessException {
	try {
		CAS cas = jcas.getCas();
		mappingProvider.configure(cas);
		
		for (Sentence sentenceAnno : JCasUtil.select(jcas, Sentence.class)) {
			List<Token> tokens = JCasUtil.selectCovered(jcas, Token.class, sentenceAnno);
			List<String> tokenStrings = JCasUtil.toText(tokens);
			List<NamedEntityWord> taggedTokens;
			
			synchronized (innerTool) {
				innerTool.setSentence(tokenStrings);
				innerTool.recognize();
				taggedTokens = innerTool.getAnnotatedSentence();
			}
			
			if (taggedTokens.size() != tokens.size()) {
				throw new NamedEntityRecognizerException("Got NER tagging for " + taggedTokens.size() +
						" tokens, should have gotten according to the total number of tokens in the sentence: " + tokens.size());
			}
			
			Iterator<Token> tokenIter = tokens.iterator();
			for (NamedEntityWord taggedToken : taggedTokens) {
				Token tokenAnno = tokenIter.next();
				
				if (taggedToken.getNamedEntity() != null) {
				
					// We rely on the fact the NamedEntity enum values have the same names as the ones
					// specified in the DKPro mapping (e.g. PERSON, ORGANIZATION)
					String tagString = taggedToken.getNamedEntity().toString();
					
					// Get an annotation with the appropriate UIMA type via the mappingProvider
					Type nerTag = mappingProvider.getTagType(tagString);
					NamedEntity nerAnnotation = (NamedEntity) cas.createAnnotation(nerTag, tokenAnno.getBegin(), tokenAnno.getEnd());
					nerAnnotation.setValue(tagString);
					nerAnnotation.addToIndexes();
				}
			}
		}
	} catch (NamedEntityRecognizerException e) {
		throw new AnalysisEngineProcessException(AnalysisEngineProcessException.ANNOTATOR_EXCEPTION, null, e);
	}
}
 

import org.uimafit.util.JCasUtil; //导入方法依赖的package包/类
private void checkAnnotations(JCas doc) {
	Collection<Second> seconds = JCasUtil.select(doc, Second.class);
	Second s1 = null;
	Second s2 = null;
	assertTrue(seconds.size() == 2);
	for (Second s : seconds) {
		if (s.getBegin() == 6) {
			assertTrue(s.getEnd() == 10);
			s1 = s;
		}
		if (s.getBegin() == 5) {
			assertTrue(s.getEnd() == 9);
			s2 = s;
		}
	}

	DocumentAnnotation docAnnot = JCasUtil.selectSingle(
		doc,
		DocumentAnnotation.class);
	assertTrue(docAnnot != null);

	Collection<First> firsts = JCasUtil.select(doc, First.class);
	assertTrue(firsts.size() == 2);

	for (First f : firsts) {
		if (f.getBegin() == 1) {
			assertTrue(f.getEnd() == 3);
			assertTrue(f.getAnnot() == s1);
			assertTrue(f.getBool());
			assertFalse(f.getBoolArray(0));
			assertTrue(f.getBoolArray(2));
			assertTrue(f.getDocAnnot() == docAnnot);
			assertTrue(f.getFirst() == f);
			assertTrue(f.getInteger() == 123);
			assertTrue(f.getIntegerList().getNthElement(0) == 5467);
			assertTrue(f.getIntegerList().getNthElement(1) == 345);
			assertTrue(f.getRefToSofa() == doc.getSofa());
			assertTrue(f.getSecond() == s2);
			assertTrue(f.getString().equals("String value"));
		}
		if (f.getBegin() == 0) {
			assertTrue(f.getFirst() != null);
		}
	}
}