@Override
public void initialize() throws ResourceInitializationException {
this.queueName = (String) getConfigParameterValue(PARAM_QUEUE_NAME);
this.mLanguage = Lang.forName((String) getConfigParameterValue(PARAM_LANGUAGE));
this.streamName = (String) getConfigParameterValue(PARAM_NAME);
documentQueue = QueueRegistry.getInstance().getQueue(queueName);
this.mCurrentIndex = 0;
this.cumulatedLength = 0;
}
@Override
public void getNext(CAS cas) throws IOException, CollectionException {
this.cumulatedLength += currentDoc.getText().length();
logger.info(
"[Stream {}] Processing document {}: {} (total length processed: {})",
this.streamName,
this.mCurrentIndex,
this.currentDoc.getUri(),
this.cumulatedLength);
SourceDocumentInformation sdi;
try {
sdi = new SourceDocumentInformation(cas.getJCas());
sdi.setUri(currentDoc.getUri());
cas.setDocumentLanguage(mLanguage.getCode());
cas.setDocumentText(currentDoc.getText());
sdi.setDocumentSize(currentDoc.getText().length());
sdi.setCumulatedDocumentSize(this.cumulatedLength);
sdi.setBegin(0);
sdi.setEnd(currentDoc.getText().length());
sdi.setOffsetInSource(0);
sdi.setDocumentIndex(mCurrentIndex);
/*
* Cannot be known in case of streaming
*/
sdi.setCorpusSize(-1);
sdi.setNbDocuments(-1);
// Cannot know if this is the last
sdi.setLastSegment(false);
sdi.addToIndexes();
this.mCurrentIndex++;
} catch (CASException e) {
throw new CollectionException(e);
}
}
