Java Code Examples for org.apache.uima.jcas.JCas#reset()

@Test
public void testTearline() throws Exception {
  JCas jCas = JCasSingleton.getJCasInstance();

  BaleenContentExtractor contentExtractor = new TearlineContentExtractor();
  contentExtractor.initialize(new CustomResourceSpecifier_impl(), Collections.emptyMap());

  String[] files = new String[] {"1.docx", "2.docx", "3.docx", "4.docx", "5.doc", "6.pdf"};
  for (String file : files) {
    File f = new File(getClass().getResource("tearline/" + file).getPath());

    try (InputStream is = new FileInputStream(f); ) {
      contentExtractor.processStream(is, f.getPath(), jCas);
      assertEquals("This is the first tearline.", jCas.getDocumentText());

      jCas.reset();
    }
  }
  contentExtractor.destroy();
}
 

@Test
public void testNoTearline() throws Exception {
  JCas jCas = JCasSingleton.getJCasInstance();

  BaleenContentExtractor contentExtractor = new TearlineContentExtractor();
  contentExtractor.initialize(new CustomResourceSpecifier_impl(), Collections.emptyMap());

  File f = new File(getClass().getResource("tearline/notearline.docx").getPath());

  try (InputStream is = new FileInputStream(f); ) {
    contentExtractor.processStream(is, f.getPath(), jCas);
    assertEquals("This document has no tearline.", jCas.getDocumentText());

    jCas.reset();
  }
  contentExtractor.destroy();
}
 

@Test
public void testBoilerplate() throws Exception {
  JCas jCas = JCasSingleton.getJCasInstance();

  Map<String, Object> params = new HashMap<>();
  params.put("boilerplate", new String[] {"[aeiou]"});

  BaleenContentExtractor contentExtractor = new TearlineContentExtractor();
  contentExtractor.initialize(new CustomResourceSpecifier_impl(), params);

  File f = new File(getClass().getResource("tearline/notearline.docx").getPath());

  try (InputStream is = new FileInputStream(f); ) {
    contentExtractor.processStream(is, f.getPath(), jCas);
    assertEquals("Ths dcmnt hs n trln.", jCas.getDocumentText());

    jCas.reset();
  }
  contentExtractor.destroy();
}
 

@Test
public void testCustomTearline() throws Exception {
  JCas jCas = JCasSingleton.getJCasInstance();

  Map<String, Object> params = new HashMap<>();
  params.put("tearline", "Customer Form:");

  BaleenContentExtractor contentExtractor = new TearlineContentExtractor();
  contentExtractor.initialize(new CustomResourceSpecifier_impl(), params);

  File f = new File(getClass().getResource("tearline/customtearline.docx").getPath());

  try (InputStream is = new FileInputStream(f); ) {
    contentExtractor.processStream(is, f.getPath(), jCas);
    assertEquals("This is the first tearline.", jCas.getDocumentText());

    jCas.reset();
  }
  contentExtractor.destroy();
}
 

public void initCas(JCas jcas) {

		jcas.reset();
		jcas.setDocumentText(text);

		Sentence s = new Sentence(jcas);
		s.setBegin(0);
		s.setEnd(text.length());
		s.addToIndexes(jcas);

		String[] tokens = text.split(" ");
		int j = 0;
		for (int i = 0; i < tokens.length; i++) {
			Token token = new Token(jcas);
			token.setBegin(j);
			token.setEnd(j + tokens[i].length());
			j = j + tokens[i].length() + 1;
			token.addToIndexes(jcas);
		}
	}
 

/**
 * Checks in a JCas to the pool. This automatically calls the {@link JCas#reset()} method, to
 * ensure that when the JCas is later retrieved from the pool it will be ready to use. Also
 * notifies other Threads that may be waiting for an instance to become available.
 * 
 * @param aJCas
 *          the JCas to release
 */
public synchronized void releaseJCas(JCas aJCas) {
  // make sure this CAS actually belongs to this pool and is checked out
  if (!mAllInstances.contains(aJCas) || mFreeInstances.contains(aJCas)) {
    UIMAFramework.getLogger(CLASS_NAME).logrb(Level.WARNING, CLASS_NAME.getName(), "releaseJCas",
            LOG_RESOURCE_BUNDLE, "UIMA_return_jcas_to_pool__WARNING");
  } else {
    // reset CAS
    aJCas.reset();
    // Add the CAS to the end of the free instances List
    mFreeInstances.add(aJCas);
  }

  // Notify any threads waiting on this object
  notifyAll();
}
 

private void createDocument(JCas jCas) {
  jCas.reset();
  jCas.setDocumentText("Simon was born in November 1980 in London, UK");

  Person p = new Person(jCas);
  p.setValue("Simon");
  p.setBegin(0);
  p.setEnd(5);
  p.addToIndexes();

  Temporal d = new Temporal(jCas);
  d.setValue(NOVEMBER);
  d.setBegin(18);
  d.setEnd(26);
  d.addToIndexes();

  Location l = new Location(jCas);
  l.setValue(LONDON);
  l.setBegin(35);
  l.setEnd(41);
  l.addToIndexes();

  // entity value different to covering text "UK"
  Location l2 = new Location(jCas);
  l2.setValue(UNITED_KINGDOM);
  l2.setBegin(43);
  l2.setEnd(45);
  l2.addToIndexes();

  assertEquals(1, JCasUtil.select(jCas, Person.class).size());
  assertEquals(1, JCasUtil.select(jCas, Temporal.class).size());
  assertEquals(2, JCasUtil.select(jCas, Location.class).size());
}
 

@Test
public void testAfterClearDifferentUUID() throws UIMAException {

  final SharedIdGenerator generator = createSharedIdGenerator();

  final JCas jCas = JCasSingleton.getJCasInstance();
  jCas.setDocumentText("test");

  final String p1id = "1234";
  final String firstUUID = generator.generateForExternalId(p1id);

  assertEquals(firstUUID, generator.generateForExternalId(p1id));

  jCas.reset();
  jCas.setDocumentText("next");
  generator.resetIfNewJCas(jCas);

  final String secondUUID = generator.generateForExternalId(p1id);

  assertNotEquals(firstUUID, secondUUID);

  assertEquals(secondUUID, generator.generateForExternalId(p1id));
}