@Override
public synchronized String nextSentence() {
if (sentences == null || !sentences.hasNext()) {
try {
if (getReader().hasNext()) {
CAS cas = resource.retrieve();
try {
getReader().getNext(cas);
} catch (Exception e) {
log.warn("Done iterating returning an empty string");
return "";
}
resource.getAnalysisEngine().process(cas);
List<String> list = new ArrayList<>();
for (Sentence sentence : JCasUtil.select(cas.getJCas(), Sentence.class)) {
list.add(sentence.getCoveredText());
}
sentences = list.iterator();
// needs to be next cas
while (!sentences.hasNext()) {
// sentence is empty; go to another cas
if (reader.hasNext()) {
cas.reset();
getReader().getNext(cas);
resource.getAnalysisEngine().process(cas);
for (Sentence sentence : JCasUtil.select(cas.getJCas(), Sentence.class)) {
list.add(sentence.getCoveredText());
}
sentences = list.iterator();
} else return null;
}
String ret = sentences.next();
if (this.getPreProcessor() != null) ret = this.getPreProcessor().preProcess(ret);
return ret;
}
return null;
} catch (Exception e) {
throw new RuntimeException(e);
}
} else {
String ret = sentences.next();
if (this.getPreProcessor() != null) ret = this.getPreProcessor().preProcess(ret);
return ret;
}
}
public static void main(String[] args) throws Exception {
String sLine;
long startTime = System.currentTimeMillis();
URL descUrl =
VectorSpaceRetrieval.class.getResource(
"/descriptors/retrievalsystem/VectorSpaceRetrieval.xml");
if (descUrl == null) {
throw new IllegalArgumentException("Error opening VectorSpaceRetrieval.xml");
}
// create AnalysisEngine
XMLInputSource input = new XMLInputSource(descUrl);
AnalysisEngineDescription desc =
UIMAFramework.getXMLParser().parseAnalysisEngineDescription(input);
AnalysisEngine anAnalysisEngine = UIMAFramework.produceAnalysisEngine(desc);
CAS aCas = anAnalysisEngine.newCAS();
URL docUrl = VectorSpaceRetrieval.class.getResource("/data/documents.txt");
if (docUrl == null) {
throw new IllegalArgumentException("Error opening data/documents.txt");
}
BufferedReader br = new BufferedReader(new InputStreamReader(docUrl.openStream()));
while ((sLine = br.readLine()) != null) {
aCas.setDocumentText(sLine);
anAnalysisEngine.process(aCas);
aCas.reset();
}
br.close();
br = null;
anAnalysisEngine.collectionProcessComplete();
anAnalysisEngine.destroy();
long endTime = System.currentTimeMillis();
double totalTime = (endTime - startTime) / 1000.0;
System.out.println("Total time taken: " + totalTime);
}