public void getNext(JCas jCas) throws IOException, CollectionException {
String[] split = nextLine.split("\t");
checkEquals(3, split.length, "pmid" + split[0]);
String pmid = unescapeCsv(split[0]).replaceAll("\"", ""),
title = unescapeCsv(split[1]),
txt = unescapeCsv(split[2]);
Header header = new Header(jCas);
header.setDocId(pmid);
header.setTitle(title);
header.addToIndexes();
jCas.setDocumentText(title + (title.endsWith(".") ? " " : ". ") + txt);
jCas.setDocumentLanguage("en");
}
@BeforeClass
public static void before() throws Exception {
conn = getTestConn("MongoTest");
// write one test document into Mongo
JCas jCas = UimaTests.getTokenizedTestCas();
BiolexiconDictTerm b = new BiolexiconDictTerm(jCas, 0, 11);
b.setEntityId("theId");
b.addToIndexes();
Header h = new Header(jCas);
h.setDocId("17");
h.addToIndexes();
JcasPipelineBuilder p = new JcasPipelineBuilder(jCas);
p.add(MongoWriter.class, BlueUima.PARAM_DB_CONNECTION, conn);
p.process();
}
@Test
public void testRead() throws Exception {
// read
List<JCas> l =
asList(createReader(MongoCollectionReader.class, BlueUima.PARAM_DB_CONNECTION, conn));
assertEquals(1, l.size());
JCas jCas = l.get(0);
assertEquals(UimaTests.TEST_SENTENCE, jCas.getDocumentText());
for (Annotation a : JCasUtil.select(jCas, Annotation.class)) {
System.out.println(a);
}
BiolexiconDictTerm b = selectSingle(jCas, BiolexiconDictTerm.class);
assertNotNull(b);
assertEquals("theId", b.getEntityId());
Header h = selectSingle(jCas, Header.class);
assertNotNull(h);
assertEquals("17", h.getDocId());
}
@Test
public void testFormat() throws Exception {
File tmp = File.createTempFile("FromFilelistReaderTest2", null);
TextFileWriter writer = new TextFileWriter(tmp);
writer.addLine(BLUE_UTILS_TEST_BASE + "testData/1.pdf");
writer.close();
List<JCas> jCases =
asList(
createReader(
FromFilelistReader.class,
PARAM_INPUT_FILE,
tmp.getAbsolutePath(),
BlueUima.PARAM_FORMAT,
true));
assertEquals(1, jCases.size());
assertEquals(1, getHeaderIntDocId(jCases.get(0)));
Header header = selectSingle(jCases.get(0), Header.class);
assertEquals(BLUE_UTILS_TEST_BASE + "testData/1.pdf", header.getSource());
}
