@Override
protected void process(Document doc, String fieldName, List<AnnotatedField> fieldValues)
throws PipelineException {
for (AnnotatedField fieldValue : fieldValues) {
messageDigest.update(fieldValue.getValue().getBytes(CHARSET));
}
}
private static void process(String fieldName, AnnotatedField field) {
final String text = field.getValue();
if (text == null) {
log.debug("Field '{}' - null", fieldName);
} else {
List<Annotation> annotations = new ArrayList<Annotation>();
int from = -1;
for (int i = 0; i < text.length(); i++) {
final char c = text.charAt(i);
if (from < 0) {
if (!isWhiteSpace(c)) {
from = i;
} else {
continue;
}
}
if (Arrays.binarySearch(CHARS_SENT, c) >= 0 && from < i - 1) {
if (i < text.length() - 1 && isWhiteSpace(text.charAt(i + 1))) {
annotations.add(new BaseAnnotation(from, i));
from = -1;
}
}
}
if (from >= 0 && from < text.length() - 1) {
annotations.add(new BaseAnnotation(from, text.length()));
}
log.debug("Field '{}' sentence annotations: {}", fieldName, annotations.size());
field.add(SENTENCE, annotations);
}
}
