Java Code Examples for org.apache.uima.fit.util.JCasUtil#selectByIndex()

@Test
public void testExistingReferentsNoMerge() throws Exception {
  AnalysisEngine ae = AnalysisEngineFactory.createEngine(CorefBrackets.class);

  populateJCasMergeTest(jCas);
  ae.process(jCas);

  assertEquals(2, JCasUtil.select(jCas, ReferenceTarget.class).size());

  Location l = JCasUtil.selectByIndex(jCas, Location.class, 0);
  Coordinate c = JCasUtil.selectByIndex(jCas, Coordinate.class, 0);

  ReferenceTarget lRt = l.getReferent();
  ReferenceTarget cRt = c.getReferent();

  assertNotEquals(lRt, cRt);
}
 

@Test
public void testmultipleHitsWithText() throws Exception {

  AnalysisEngineDescription aed =
      AnalysisEngineFactory.createEngineDescription(
          List.class, List.PARAM_TERMS, terms, List.PARAM_TYPE, LOCATION);

  AnalysisEngine ae = AnalysisEngineFactory.createEngine(aed);

  // the same search term appears multiple times in text...
  jCas.setDocumentText("Hello world, and hello world again.");
  // but then subset using a Text annotation
  new Text(jCas, 10, jCas.getDocumentText().length()).addToIndexes();

  ae.process(jCas);

  assertEquals(1, JCasUtil.select(jCas, Location.class).size());
  Location l = JCasUtil.selectByIndex(jCas, Location.class, 0);
  assertEquals(WORLD, l.getValue());
  assertEquals(WORLD, l.getCoveredText());
  assertTrue(l.getBegin() > 10);

  ae.destroy();
}
 

@Test
public void testExistingLocReferent() throws Exception {
  AnalysisEngine ae = AnalysisEngineFactory.createEngine(CorefBrackets.class);

  jCas.setDocumentText(LOC_TEXT);

  ReferenceTarget rt1 = Annotations.createReferenceTarget(jCas);

  Location l1 = Annotations.createLocation(jCas, 0, 9, SOMEWHERE, null);
  l1.setReferent(rt1);
  Annotations.createCoordinate(jCas, 11, 19, MRGS);

  ae.process(jCas);

  assertEquals(1, JCasUtil.select(jCas, ReferenceTarget.class).size());

  ReferenceTarget rt = JCasUtil.selectByIndex(jCas, ReferenceTarget.class, 0);
  Location l = JCasUtil.selectByIndex(jCas, Location.class, 0);
  Coordinate c = JCasUtil.selectByIndex(jCas, Coordinate.class, 0);

  assertEquals(rt, l.getReferent());
  assertEquals(rt, c.getReferent());
  assertEquals(l.getReferent(), c.getReferent());
}
 

@Test
public void testMultiple() throws Exception {
  jCas.setDocumentText("His name was Andrew Smith (Drew) (Smithy)");

  Person p = new Person(jCas, 13, 41);
  p.addToIndexes();

  processJCas();

  assertEquals(3, JCasUtil.select(jCas, Person.class).size());

  Person p1 = JCasUtil.selectByIndex(jCas, Person.class, 0);
  assertEquals("Andrew Smith", p1.getCoveredText());

  Person p2 = JCasUtil.selectByIndex(jCas, Person.class, 1);
  assertEquals("Drew", p2.getCoveredText());

  Person p3 = JCasUtil.selectByIndex(jCas, Person.class, 2);
  assertEquals("Smithy", p3.getCoveredText());

  assertEquals(p1.getReferent(), p2.getReferent());
  assertEquals(p1.getReferent(), p3.getReferent());
}
 

@Test
public void testmultipleHits() throws Exception {

  AnalysisEngineDescription aed =
      AnalysisEngineFactory.createEngineDescription(
          List.class, List.PARAM_TERMS, terms, List.PARAM_TYPE, LOCATION);

  AnalysisEngine ae = AnalysisEngineFactory.createEngine(aed);

  // the same search term appears multiple times in text...
  jCas.setDocumentText("Hello world, and hello world again.");

  ae.process(jCas);

  assertEquals(2, JCasUtil.select(jCas, Location.class).size());
  Location l = JCasUtil.selectByIndex(jCas, Location.class, 0);
  assertEquals(WORLD, l.getValue());
  assertEquals(WORLD, l.getCoveredText());

  ae.destroy();
}
 

@Test
public void testWrongType() throws Exception {
  jCas.setDocumentText("It was midnight, and all was quiet");

  Temporal t = new Temporal(jCas, 7, 15);
  t.setTimestampStart(LocalDateTime.of(2016, 10, 4, 0, 0, 0).toEpochSecond(ZoneOffset.UTC));
  t.setTimestampStop(LocalDateTime.of(2016, 10, 4, 0, 0, 0).toEpochSecond(ZoneOffset.UTC));
  t.setScope("SINGLE");
  t.setTemporalType("TIME");
  t.setValue("midnight");
  t.addToIndexes();

  processJCas(
      NormalizeTemporal.PARAM_DATE_FORMAT,
      "HH:mm",
      NormalizeTemporal.PARAM_TEMPORAL_TYPE,
      "DATE");

  assertEquals(1, JCasUtil.select(jCas, Temporal.class).size());
  Temporal tTest = JCasUtil.selectByIndex(jCas, Temporal.class, 0);
  assertEquals("midnight", tTest.getValue());
  assertFalse(tTest.getIsNormalised());
}
 

@Test
public void test() throws Exception {
  AnalysisEngine rneAE = AnalysisEngineFactory.createEngine(RemoveNestedEntities.class);

  populateJCas(jCas);
  rneAE.process(jCas);

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

  Temporal dt = JCasUtil.selectByIndex(jCas, Temporal.class, 0);
  assertEquals("December 1972", dt.getCoveredText());

  Location l = JCasUtil.selectByIndex(jCas, Location.class, 0);
  assertEquals("Oxford", l.getCoveredText());
}
 

@Test
public void test() throws Exception {
  jCas.setDocumentText(
      "James is a BRITISH national. Last month, he met an Irish bloke in the pub. He is friends with Bob, who is an Spanish.");
  ae.process(jCas);

  assertEquals(3, JCasUtil.select(jCas, Nationality.class).size());

  Nationality british = JCasUtil.selectByIndex(jCas, Nationality.class, 0);
  assertNotNull(british);
  assertEquals("BRITISH", british.getCoveredText());
  assertEquals("BRITISH", british.getValue());
  assertEquals("GBR", british.getCountryCode());

  Nationality irish = JCasUtil.selectByIndex(jCas, Nationality.class, 1);
  assertNotNull(irish);
  assertEquals("Irish", irish.getCoveredText());
  assertEquals("Irish", irish.getValue());
  assertEquals("IRL", irish.getCountryCode());

  Nationality spanish = JCasUtil.selectByIndex(jCas, Nationality.class, 2);
  assertNotNull(spanish);
  assertEquals("Spanish", spanish.getCoveredText());
  assertEquals("Spanish", spanish.getValue());
  assertEquals("ESP", spanish.getCountryCode());
}
 

@Test
public void testNoExistingReferents() throws Exception {
  AnalysisEngine ae = AnalysisEngineFactory.createEngine(CorefBrackets.class);

  jCas.setDocumentText(LOC_TEXT);

  Annotations.createLocation(jCas, 0, 9, SOMEWHERE, null);
  Annotations.createCoordinate(jCas, 11, 19, MRGS);

  ae.process(jCas);

  assertEquals(1, JCasUtil.select(jCas, ReferenceTarget.class).size());

  ReferenceTarget rt = JCasUtil.selectByIndex(jCas, ReferenceTarget.class, 0);
  Location l = JCasUtil.selectByIndex(jCas, Location.class, 0);
  Coordinate c = JCasUtil.selectByIndex(jCas, Coordinate.class, 0);

  assertEquals(rt, l.getReferent());
  assertEquals(rt, c.getReferent());
}
 

@Test
public void testNested() throws Exception {
  jCas.setDocumentText("A statement US Army Major J Bloggs said that");

  Organisation o = new Organisation(jCas, 12, 34);
  o.addToIndexes();

  Person p = new Person(jCas, 20, 34);
  p.addToIndexes();

  processJCas();

  assertEquals(1, JCasUtil.select(jCas, Relation.class).size());
  Relation r = JCasUtil.selectByIndex(jCas, Relation.class, 0);
  assertEquals("US Army Major J Bloggs", r.getCoveredText());
  assertEquals("ROLE", r.getRelationshipType());

  assertEquals(1, JCasUtil.select(jCas, Organisation.class).size());
  Organisation org = JCasUtil.selectByIndex(jCas, Organisation.class, 0);
  assertEquals("US Army", org.getCoveredText());
}
 

@Test
public void testMissingType() throws Exception {
  AnalysisEngine regexAE =
      AnalysisEngineFactory.createEngine(
          Custom.class, Custom.PARAM_PATTERN, DIGIT_REGEX, Custom.PARAM_CASE_SENSITIVE, true);

  jCas.setDocumentText(TEXT);
  regexAE.process(jCas);

  assertEquals(1, JCasUtil.select(jCas, Entity.class).size());

  Entity e1 = JCasUtil.selectByIndex(jCas, Entity.class, 0);
  assertNotNull(e1);
  assertEquals(P123, e1.getCoveredText());
  assertEquals(P123, e1.getValue());

  regexAE.destroy();
}
 

@Test
public void testCaseInsensitive() throws Exception {
  AnalysisEngine regexAE =
      AnalysisEngineFactory.createEngine(
          Custom.class,
          Custom.PARAM_TYPE,
          UK_GOV_DSTL_BALEEN_TYPES_COMMON_PERSON,
          Custom.PARAM_PATTERN,
          DIGIT_REGEX,
          Custom.PARAM_CASE_SENSITIVE,
          false);

  jCas.setDocumentText(TEXT);
  regexAE.process(jCas);

  assertEquals(2, JCasUtil.select(jCas, Person.class).size());

  Person p1 = JCasUtil.selectByIndex(jCas, Person.class, 0);
  assertNotNull(p1);
  assertEquals(P123, p1.getCoveredText());
  assertEquals(P123, p1.getValue());

  Person p2 = JCasUtil.selectByIndex(jCas, Person.class, 1);
  assertNotNull(p2);
  assertEquals(P456, p2.getCoveredText());
  assertEquals(P456, p2.getValue());

  regexAE.destroy();
}
 

@Test
public void testFull() throws Exception {
  jCas.setDocumentText(
      "Here's a full IPv6 address fe80:0000:0000:0000:0204:61ff:fe9d:f156 and some text after it");
  processJCas();

  assertEquals(1, JCasUtil.select(jCas, CommsIdentifier.class).size());
  CommsIdentifier ip = JCasUtil.selectByIndex(jCas, CommsIdentifier.class, 0);
  assertEquals("fe80:0000:0000:0000:0204:61ff:fe9d:f156", ip.getCoveredText());
  assertEquals("ipv6address", ip.getSubType());
}
 

@Test
public void test() throws Exception {
  jCas.setDocumentText("Forty seven knights conspired against the crown.");
  processJCas(MONGO, erd, COLLECTION, MONGO_COLL, TYPE, BUZZWORD);

  assertEquals(2, JCasUtil.select(jCas, Buzzword.class).size());

  Buzzword b1 = JCasUtil.selectByIndex(jCas, Buzzword.class, 0);
  assertEquals("knights", b1.getValue());
  assertEquals("knights", b1.getCoveredText());

  Buzzword b2 = JCasUtil.selectByIndex(jCas, Buzzword.class, 1);
  assertEquals("conspired", b2.getValue());
  assertEquals("conspired", b2.getCoveredText());
}
 

@Test
public void testMonthDayTime() throws Exception {
  jCas.setDocumentText("It happened at Apr 22, 2014 1529Z");
  processJCas();

  assertEquals(1, JCasUtil.select(jCas, Temporal.class).size());

  Temporal ts1 = JCasUtil.selectByIndex(jCas, Temporal.class, 0);
  assertEquals("Apr 22, 2014 1529Z", ts1.getCoveredText());
  assertEquals(1398180540L, ts1.getTimestampStart());
  assertEquals(1398180600L, ts1.getTimestampStop());

  jCas.reset();

  jCas.setDocumentText("It happened at April 22 2014 1529 EST");
  processJCas();

  assertEquals(1, JCasUtil.select(jCas, Temporal.class).size());

  Temporal ts2 = JCasUtil.selectByIndex(jCas, Temporal.class, 0);
  assertEquals("April 22 2014 1529 EST", ts2.getCoveredText());
  assertEquals(1398198540L, ts2.getTimestampStart());
  assertEquals(1398198600L, ts2.getTimestampStop());

  jCas.reset();

  jCas.setDocumentText("It happened at April 22, 2014 152947Z");
  processJCas();

  assertEquals(1, JCasUtil.select(jCas, Temporal.class).size());

  Temporal ts3 = JCasUtil.selectByIndex(jCas, Temporal.class, 0);
  assertEquals("April 22, 2014 152947Z", ts3.getCoveredText());
  assertEquals(1398180587L, ts3.getTimestampStart());
  assertEquals(1398180588L, ts3.getTimestampStop());
}
 

@Test
public void testDayMonthYears() throws Exception {
  jCas.setDocumentText(
      "He is on duty from 3-10 October 2016, whilst she was on duty 27th September - Monday 3 Oct 16. The Christmas break fell between 21st December 2016 and 2 January 17. On 2/3 January '17 there was a storm, and it rained on 2nd and 5th January 2017.");
  processJCas();

  assertEquals(6, JCasUtil.select(jCas, Temporal.class).size());

  Temporal ts1 = JCasUtil.selectByIndex(jCas, Temporal.class, 0);
  assertEquals("3-10 October 2016", ts1.getCoveredText());
  assertEquals(1475452800L, ts1.getTimestampStart());
  assertEquals(1476144000L, ts1.getTimestampStop());

  Temporal ts2 = JCasUtil.selectByIndex(jCas, Temporal.class, 1);
  assertEquals("27th September - Monday 3 Oct 16", ts2.getCoveredText());
  assertEquals(1474934400L, ts2.getTimestampStart());
  assertEquals(1475539200L, ts2.getTimestampStop());

  Temporal ts3 = JCasUtil.selectByIndex(jCas, Temporal.class, 2);
  assertEquals("21st December 2016 and 2 January 17", ts3.getCoveredText());
  assertEquals(1482278400L, ts3.getTimestampStart());
  assertEquals(1483401600L, ts3.getTimestampStop());

  Temporal ts4 = JCasUtil.selectByIndex(jCas, Temporal.class, 3);
  assertEquals("2/3 January '17", ts4.getCoveredText());
  assertEquals(1483315200L, ts4.getTimestampStart());
  assertEquals(1483488000L, ts4.getTimestampStop());

  Temporal ts5 = JCasUtil.selectByIndex(jCas, Temporal.class, 4);
  assertEquals("2nd and 5th January 2017", ts5.getCoveredText());
  assertEquals("2nd January 2017", ts5.getValue());
  assertEquals(1483315200L, ts5.getTimestampStart());
  assertEquals(1483401600L, ts5.getTimestampStop());

  Temporal ts6 = JCasUtil.selectByIndex(jCas, Temporal.class, 5);
  assertEquals("5th January 2017", ts6.getCoveredText());
  assertEquals(1483574400L, ts6.getTimestampStart());
  assertEquals(1483660800L, ts6.getTimestampStop());
}
 

@Test
public void testNormalizeLonFirst() throws Exception {
  String docText =
      "London is located in the UK, at 51.507° N, 0.125°W. Edinburgh is also in the UK, at 55.9530°N,3.1880° W."
          + "But darkest Peru is at 9.19° S, 75.0152°W and South Korea at 35.9078°N, 127.7669° E.";
  jCas.setDocumentText(docText);
  processJCas("storeDecimalValue", true, "storeLongitudeFirst", true);

  assertAnnotations(
      4,
      Coordinate.class,
      new TestCoordinate(0, "51.507° N, 0.125°W", "dd", TYPE_POINT_COORDINATES_0_125_51_507),
      new TestCoordinate(1, "55.9530°N,3.1880° W", "dd", TYPE_POINT_COORDINATES_3_188_55_953),
      new TestCoordinate(2, "9.19° S, 75.0152°W", "dd", TYPE_POINT_COORDINATES_75_0152_09_190),
      new TestCoordinate(
          3, "35.9078°N, 127.7669° E", "dd", TYPE_POINT_COORDINATES_127_766_35_907));

  Coordinate t = JCasUtil.selectByIndex(jCas, Coordinate.class, 0);
  assertEquals("-0.125 51.507", t.getValue());
  assertEquals(true, t.getIsNormalised());
  t = JCasUtil.selectByIndex(jCas, Coordinate.class, 1);
  assertEquals("-3.188 55.953", t.getValue());
  assertEquals(true, t.getIsNormalised());
  t = JCasUtil.selectByIndex(jCas, Coordinate.class, 2);
  assertEquals("-75.0152 -9.19", t.getValue());
  assertEquals(true, t.getIsNormalised());
  t = JCasUtil.selectByIndex(jCas, Coordinate.class, 3);
  assertEquals("127.7669 35.9078", t.getValue());
  assertEquals(true, t.getIsNormalised());
}
 

@Test
public void test1() throws UIMAException {
  createPOS();

  Person person = new Person(jCas, 0, 2); // He
  person.addToIndexes();

  Money money = new Money(jCas, 59, 71); // $1.8 billion
  money.addToIndexes();

  Temporal date = new Temporal(jCas, 75, 84); // September
  date.addToIndexes();

  processJCas();

  assertEquals(2, JCasUtil.select(jCas, Relation.class).size());

  Relation r1 = JCasUtil.selectByIndex(jCas, Relation.class, 0);
  assertEquals("reckons", r1.getValue());
  assertEquals(person, r1.getSource());
  assertEquals("uk.gov.dstl.baleen.types.semantic.Entity", r1.getTarget().getTypeName());
  assertEquals("the current account deficit", r1.getTarget().getCoveredText());

  Relation r2 = JCasUtil.selectByIndex(jCas, Relation.class, 1);
  assertEquals("will narrow to", r2.getValue());
  assertEquals("uk.gov.dstl.baleen.types.semantic.Entity", r2.getSource().getTypeName());
  assertEquals("the current account deficit", r2.getSource().getCoveredText());
  assertEquals(money, r2.getTarget());
}
 

@Test
public void test() throws Exception {
  File f = new File(getClass().getResource("lineReader.txt").getPath());
  BaleenCollectionReader bcr = getCollectionReader(LineReader.PARAM_FILE, f.getPath());

  assertTrue(bcr.doHasNext());
  bcr.getNext(jCas.getCas());
  assertEquals("This is the first line", jCas.getDocumentText());
  assertEquals(2, JCasUtil.select(jCas, Metadata.class).size());
  Metadata md = JCasUtil.selectByIndex(jCas, Metadata.class, 1);
  assertEquals("lineNumber", md.getKey());
  assertEquals("1", md.getValue());
  assertTrue(getSource(jCas).endsWith("#1"));
  jCas.reset();

  assertTrue(bcr.doHasNext());
  bcr.getNext(jCas.getCas());
  assertEquals("This is the second line", jCas.getDocumentText());
  assertEquals(2, JCasUtil.select(jCas, Metadata.class).size());
  md = JCasUtil.selectByIndex(jCas, Metadata.class, 1);
  assertEquals("lineNumber", md.getKey());
  assertEquals("2", md.getValue());
  assertTrue(getSource(jCas).endsWith("#2"));
  jCas.reset();

  assertTrue(bcr.doHasNext());
  bcr.getNext(jCas.getCas());
  assertEquals("This is the fourth line, but the third one we pick out", jCas.getDocumentText());
  md = JCasUtil.selectByIndex(jCas, Metadata.class, 1);
  assertEquals("lineNumber", md.getKey());
  assertEquals("4", md.getValue());
  assertTrue(getSource(jCas).endsWith("#4"));
  jCas.reset();

  assertTrue(bcr.doHasNext());
  bcr.getNext(jCas.getCas());
  assertEquals(
      "This is the sixth line, but the fourth and final one we pick out", jCas.getDocumentText());
  md = JCasUtil.selectByIndex(jCas, Metadata.class, 1);
  assertEquals("lineNumber", md.getKey());
  assertEquals("6", md.getValue());
  assertTrue(getSource(jCas).endsWith("#6"));
  jCas.reset();

  assertFalse(bcr.doHasNext());
  assertFalse(bcr.doHasNext());

  bcr.close();
}
 

@Test
public void testCoref() throws Exception {
  ExternalResourceDescription erd =
      ExternalResourceFactory.createNamedResourceDescription(
          MONGO,
          SharedFongoResource.class,
          FONGO_COLLECTION,
          MONGO_COLL,
          FONGO_DATA,
          objectMapper.writeValueAsString(GAZ_DATA));
  AnalysisEngineDescription aed =
      AnalysisEngineFactory.createEngineDescription(
          MongoRegex.class,
          MONGO,
          erd,
          COLLECTION,
          MONGO_COLL,
          TYPE,
          LOCATION,
          REGEX,
          "\\b[A-Z][a-z]*\\b");

  AnalysisEngine ae = AnalysisEngineFactory.createEngine(aed);

  jCas.setDocumentText("Hello World, Hello Earth");

  ae.process(jCas);

  assertEquals(2, JCasUtil.select(jCas, Location.class).size());
  assertEquals(1, JCasUtil.select(jCas, ReferenceTarget.class).size());

  ReferenceTarget rt = JCasUtil.selectByIndex(jCas, ReferenceTarget.class, 0);

  Location l1 = JCasUtil.selectByIndex(jCas, Location.class, 0);
  assertEquals("World", l1.getValue());
  assertEquals("World", l1.getCoveredText());
  assertEquals(rt, l1.getReferent());

  Location l2 = JCasUtil.selectByIndex(jCas, Location.class, 1);
  assertEquals("Earth", l2.getValue());
  assertEquals("Earth", l2.getCoveredText());
  assertEquals(rt, l2.getReferent());

  ae.destroy();
}