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))); }
da.setTimestamp(calculateBestDate(message, file)); da.setDocType("email"); da.setDocumentClassification("O"); String source = file.getAbsolutePath().substring(rootFolder.length()); da.setSourceUri(source);
da.setTimestamp(calculateBestDate(message, file)); da.setDocType("email"); da.setDocumentClassification("O"); da.setSourceUri(file.getAbsolutePath().substring(rootFolder.length())); da.setLanguage("en");
da.setTimestamp(System.currentTimeMillis()); da.setDocType("re3d"); da.setDocumentClassification("O"); da.setSourceUri(doc.getSourceUrl());
da.setTimestamp(System.currentTimeMillis()); da.setDocType("re3d"); da.setDocumentClassification("O"); da.setSourceUri(doc.getSourceUrl());
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; }
da.setSourceUri("test.txt"); da.setDocumentClassification("UK OFFICIAL"); da.setDocumentCaveats(UimaTypesUtils.toArray(jCas, Arrays.asList("Test", "Caveats")));
@Test public void testReindexEntities() throws Exception { createEntitiesDocument(); ae.process(jCas); ae.process(jCas); // Change the last document so we can check its been updated getDocumentAnnotation(jCas).setDocumentClassification("TEST"); ae.process(jCas); elasticsearch.flush(BALEEN_INDEX); assertEquals(new Long(1), getCount()); SearchHit result = elasticsearch.client().search(new SearchRequest()).actionGet().getHits().getHits()[0]; // This checks the last document is tone we are getting assertEquals("TEST", result.getSource().get("classification")); }
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.setDocumentCaveats(UimaTypesUtils.toArray(jCas, Arrays.asList(CAVEAT))); da.setDocumentReleasability(UimaTypesUtils.toArray(jCas, Arrays.asList(RELEASABILITY))); da.setDocumentClassification(CLASSIFICATION); da.setSourceUri(SOURCE); da.setLanguage(LANGUAGE);
public JCasSerializationTester() throws UIMAException { out = JCasFactory.createJCas(); in = JCasFactory.createJCas(); in.setDocumentText("This is a test"); final DocumentAnnotation inDA = (DocumentAnnotation) in.getDocumentAnnotationFs(); inDA.setDocumentClassification("CLASS"); final Location l = new Location(in); l.setBegin(10); l.setEnd(12); l.setGeoJson(GEO_JSON); l.setValue("Dinagat Islands"); l.setConfidence(0.9d); l.addToIndexes(in); final ReferenceTarget rt = new ReferenceTarget(in); rt.setBegin(0); rt.setEnd(1); rt.addToIndexes(in); final Person p = new Person(in); p.setBegin(2); p.setEnd(8); p.setGender("Male"); p.setValue("Peter"); p.setConfidence(0.5d); p.setReferent(rt); p.addToIndexes(in); }
da.setDocumentClassification("CLASS"); da.setDocType("MANUAL"); da.setSourceUri("http://test.com");