public static String getTermSuiteCasFileName(JCas jcas) {
FSIterator<Annotation> it = jcas.getAnnotationIndex(SourceDocumentInformation.type).iterator();
if (it.hasNext()) {
SourceDocumentInformation sdi = (SourceDocumentInformation) it.next();
Iterator<String> iterator =
Splitter.on("/").split(sdi.getUri() == null ? "(no uri)" : sdi.getUri()).iterator();
String name = null;
while (iterator.hasNext()) name = iterator.next();
return name;
} else return null;
}
public static void initJCasSDI(JCas jCas, String language, String text, String uri) {
SourceDocumentInformation sdi;
sdi = new SourceDocumentInformation(jCas);
sdi.setBegin(0);
sdi.setUri(uri);
sdi.setEnd(text.length());
sdi.setOffsetInSource(0);
sdi.addToIndexes();
}
@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);
}
}
