Java Code Examples for org.apache.uima.jcas.tcas.DocumentAnnotation#setDocumentCaveats()

protected long createNoEntitiesDocument() {
  jCas.reset();
  jCas.setDocumentText("Hello World");
  jCas.setDocumentLanguage("en");

  long timestamp = System.currentTimeMillis();

  DocumentAnnotation da = getDocumentAnnotation(jCas);
  da.setTimestamp(timestamp);
  da.setSourceUri("test/no_entities");
  da.setDocType("test");
  da.setDocumentClassification("OFFICIAL");
  da.setDocumentCaveats(
      UimaTypesUtils.toArray(jCas, Arrays.asList(new String[] {"TEST_A", "TEST_B"})));
  da.setDocumentReleasability(
      UimaTypesUtils.toArray(jCas, Arrays.asList(new String[] {"ENG", "SCO", "WAL"})));

  return timestamp;
}
 

private void processDocumentAnnotation(
    final JCas jCas, final DocumentAnnotation da, final Map<String, Object> map) {
  da.setDocType((String) map.getOrDefault(JsonJCas.DA_DOCUMENT_TYPE, ""));
  da.setDocumentClassification((String) map.getOrDefault(JsonJCas.DA_CLASSIFICATION, ""));
  da.setLanguage((String) map.getOrDefault(JsonJCas.DA_LANGUAGE, ""));
  da.setSourceUri((String) map.getOrDefault(JsonJCas.DA_SOURCE_URI, ""));
  da.setTimestamp(((Number) map.getOrDefault(JsonJCas.DA_TIMESTAMP, 0)).longValue());

  da.setDocumentCaveats(
      UimaTypesUtils.toArray(
          jCas, (Collection<String>) map.getOrDefault(JsonJCas.DA_CAVEATS, null)));
  da.setDocumentReleasability(
      UimaTypesUtils.toArray(
          jCas, (Collection<String>) map.getOrDefault(JsonJCas.DA_RELEASABILITY, null)));
}
 

@Test
public void test() throws Exception {
  jCas.setDocumentText("Hello James! Is your e-mail address [email protected]? 'No', said James.");

  DocumentAnnotation da = getDocumentAnnotation(jCas);
  da.setDocType("test");
  da.setSourceUri("http://www.example.com/hello.txt");

  StringArray sa = new StringArray(jCas, 2);
  sa.set(0, "UK");
  sa.set(1, "US");
  da.setDocumentCaveats(sa);

  ReferenceTarget rt = new ReferenceTarget(jCas);
  rt.addToIndexes();

  Entity e1 = new Person(jCas, 6, 11);
  e1.setValue("James");
  e1.setReferent(rt);
  e1.addToIndexes();

  Entity e1a = new Person(jCas, 60, 65);
  e1a.setValue("James");
  e1a.setReferent(rt);
  e1a.addToIndexes();

  Entity e2 = new CommsIdentifier(jCas, 36, 47);
  e2.setValue("[email protected]");
  e2.addToIndexes();

  processJCas(GremlinConsumer.PARAM_GRAPH_CONFIG, tmpConfig.getPath());

  // TODO: Write some proper tests that actually check something
}
 

public static void populateJcas(final JCas jCas) {

    jCas.setDocumentText(CONTENT);
    final DocumentAnnotation da = (DocumentAnnotation) jCas.getDocumentAnnotationFs();
    da.setDocumentClassification("CLASS");
    da.setDocType("MANUAL");
    da.setSourceUri("http://test.com");
    da.setLanguage("en");
    da.setTimestamp(new Date().getTime());
    da.setDocumentCaveats(new StringArray(jCas, 2));
    da.setDocumentCaveats(0, "GITHUB");
    da.setDocumentCaveats(1, "CAVEAT");

    final Metadata m1 = new Metadata(jCas);
    m1.setKey("test");
    m1.setValue("1");
    m1.addToIndexes(jCas);

    final Metadata m2 = new Metadata(jCas);
    m2.setKey("test");
    m2.setValue("2");
    m2.addToIndexes(jCas);

    final PublishedId pId = new PublishedId(jCas);
    pId.setPublishedIdType("test");
    pId.setValue("12");
    pId.addToIndexes(jCas);

    ReferenceTarget target = new ReferenceTarget(jCas);
    target.setLinking("testLinking");
    target.addToIndexes(jCas);

    final Person js = new Person(jCas);
    js.setBegin(25);
    js.setEnd(35);
    js.setGender("Male");
    js.setValue("John Smith");
    js.setConfidence(0.9d);
    js.setReferent(target);
    js.addToIndexes(jCas);

    final Person jd = new Person(jCas);
    jd.setBegin(50);
    jd.setEnd(58);
    jd.setGender("Female");
    jd.setValue("Jane Doe");
    jd.setConfidence(0.8d);
    jd.addToIndexes(jCas);

    final Person he = new Person(jCas);
    he.setBegin(60);
    he.setEnd(62);
    he.setGender("Male");
    he.setValue("He");
    he.setConfidence(0.9d);
    he.setReferent(target);
    he.addToIndexes(jCas);

    final Location l = new Location(jCas);
    l.setBegin(72);
    l.setEnd(87);
    l.setGeoJson(GEO_JSON);
    l.setValue("Dinagat Islands");
    l.setConfidence(0.9d);
    l.addToIndexes(jCas);

    final Relation related = new Relation(jCas);
    related.setBegin(36);
    related.setEnd(49);
    related.setValue("is related to");
    related.setRelationshipType(RELATED_TYPE);
    related.setSource(js);
    related.setTarget(jd);
    related.addToIndexes(jCas);

    final Relation lives = new Relation(jCas);
    lives.setBegin(63);
    lives.setEnd(71);
    lives.setValue("lives at");
    lives.setRelationshipType(LIVES_TYPE);
    lives.setSource(js);
    lives.setTarget(l);
    lives.addToIndexes(jCas);

    final Event event = new Event(jCas);
    event.setBegin(0);
    event.setEnd(10);
    event.setValue("test event");
    event.setEventType(new StringArray(jCas, 1));
    event.setEventType(0, "MEETING");
    event.setEntities(new FSArray(jCas, 2));
    event.setEntities(0, js);
    event.setEntities(1, jd);
    event.setArguments(new StringArray(jCas, 2));
    event.setArguments(0, "argument");
    event.setArguments(1, "Other");
    event.addToIndexes(jCas);
  }
 

@SuppressWarnings("unchecked")
@Test
public void testNoEntities() throws Exception {
  jCas.setDocumentText(TEXT);
  jCas.setDocumentLanguage("en");

  long timestamp = System.currentTimeMillis();

  DocumentAnnotation da = getDocumentAnnotation(jCas);
  da.setTimestamp(timestamp);
  da.setSourceUri("test/no_entities");
  da.setDocType("test");
  da.setDocumentClassification("OFFICIAL");
  da.setDocumentCaveats(
      UimaTypesUtils.toArray(jCas, Arrays.asList(new String[] {"TEST_A", "TEST_B"})));
  da.setDocumentReleasability(
      UimaTypesUtils.toArray(jCas, Arrays.asList(new String[] {"ENG", "SCO", "WAL"})));

  ae.process(jCas);

  assertEquals(1, documents.count());
  Document result = documents.find().first();

  assertEquals(TEXT, result.get(Mongo.FIELD_CONTENT));
  assertEquals(
      "en", ((Document) result.get(Mongo.FIELD_DOCUMENT)).get(Mongo.FIELD_DOCUMENT_LANGUAGE));

  assertEquals(
      new Date(timestamp),
      ((Document) result.get(Mongo.FIELD_DOCUMENT)).get(Mongo.FIELD_DOCUMENT_TIMESTAMP));
  assertEquals(
      "test/no_entities",
      ((Document) result.get(Mongo.FIELD_DOCUMENT)).get(Mongo.FIELD_DOCUMENT_SOURCE));

  assertEquals(
      "test", ((Document) result.get(Mongo.FIELD_DOCUMENT)).get(Mongo.FIELD_DOCUMENT_TYPE));

  assertEquals(
      "OFFICIAL",
      ((Document) result.get(Mongo.FIELD_DOCUMENT)).get(Mongo.FIELD_DOCUMENT_CLASSIFICATION));
  assertArrayEquals(
      new String[] {"TEST_A", "TEST_B"},
      ((Collection<String>)
              ((Document) result.get(Mongo.FIELD_DOCUMENT)).get(Mongo.FIELD_DOCUMENT_CAVEATS))
          .toArray());
  assertArrayEquals(
      new String[] {"ENG", "SCO", "WAL"},
      ((Collection<String>)
              ((Document) result.get(Mongo.FIELD_DOCUMENT))
                  .get(Mongo.FIELD_DOCUMENT_RELEASABILITY))
          .toArray());

  assertEquals(getDocumentAnnotation(jCas).getHash(), result.get(fields.getExternalId()));
}