@Override
public void initialize() throws ResourceInitializationException {
super.initialize();
logger = getLogger();
String fileName = (String) getConfigParameterValue(INPUT_FILE_PARAM);
try {
file = new File(URI.create(fileName));
} catch (Exception e) {
ExceptionHandler.logAndRethrow(logger, e);
}
if (!file.exists()) {
ExceptionHandler.logAndThrow(logger, "Input file does not exists");
}
}
@Override
public void getNext(final CAS aCAS) throws IOException, CollectionException {
TikaProcessor processor = new TikaProcessor();
try {
processor = TikaProcessor.newInstance(file);
} catch (Exception e) {
ExceptionHandler.logAndRethrow(logger, "TikaProcessor: ", e);
}
String documentText = processor.getText();
if (documentText == null || documentText.length() == 0) {
ExceptionHandler.logAndThrow(logger, "Document text is null or empty");
}
aCAS.setDocumentText(documentText);
String textLanguage = processor.getLanguage();
if (!textLanguage.contains("ru")) {
ExceptionHandler.logAndThrow(logger, "Document language is not russian");
}
aCAS.setDocumentLanguage(textLanguage);
}