@Override
public DataBag exec(Tuple input) throws IOException {
DataBag bag = (DataBag) input.get(0);
long n = 0L;
DataBag selected = bagFactory.newDefaultBag();
DataBag waiting = bagFactory.newSortedBag(new ScoredTupleComparator());
for (Tuple innerTuple : bag) {
n += (Long) innerTuple.get(0);
selected.addAll((DataBag) innerTuple.get(1));
waiting.addAll((DataBag) innerTuple.get(2));
}
long sampleSize = (long) Math.ceil(_samplingProbability * n);
long nNeeded = sampleSize - selected.size();
for (Tuple scored : waiting) {
if (nNeeded <= 0) {
break;
}
selected.add(ScoredTuple.fromIntermediateTuple(scored).getTuple());
nNeeded--;
}
return selected;
}
@Override
public Tuple exec(Tuple input) throws IOException {
DataBag bag = (DataBag) input.get(0);
DataBag selected = bagFactory.newDefaultBag();
DataBag aggWaiting = bagFactory.newSortedBag(new ScoredTupleComparator());
DataBag waiting = bagFactory.newSortedBag(new ScoredTupleComparator());
Tuple output = tupleFactory.newTuple();
long n = 0L;
for (Tuple innerTuple : bag) {
n += (Long) innerTuple.get(0);
selected.addAll((DataBag) innerTuple.get(1));
double q1 = getQ1(n, _samplingProbability);
double q2 = getQ2(n, _samplingProbability);
for (Tuple t : (DataBag) innerTuple.get(2)) {
ScoredTuple scored = ScoredTuple.fromIntermediateTuple(t);
if (scored.getScore() < q1) {
selected.add(scored.getTuple());
} else if (scored.getScore() < q2) {
aggWaiting.add(t);
} else {
break;
}
}
}
double q1 = getQ1(n, _samplingProbability);
double q2 = getQ2(n, _samplingProbability);
for (Tuple t : aggWaiting) {
ScoredTuple scored = ScoredTuple.fromIntermediateTuple(t);
if (scored.getScore() < q1) {
selected.add(scored.getTuple());
} else if (scored.getScore() < q2) {
waiting.add(t);
} else {
break;
}
}
output.append(n);
output.append(selected);
output.append(waiting);
System.err.println(
"Read "
+ n
+ " items, selected "
+ selected.size()
+ ", and wait-listed "
+ aggWaiting.size()
+ ".");
return output;
}