private VectorContainer getBatch() throws IOException {
assert fs != null;
assert path != null;
if (inputStream == null) {
inputStream = fs.open(path);
}
VectorAccessibleSerializable vas = new VectorAccessibleSerializable(allocator);
Stopwatch watch = Stopwatch.createStarted();
vas.readFromStream(inputStream);
VectorContainer c = vas.get();
if (schema != null) {
c = SchemaUtil.coerceContainer(c, schema, context);
}
// logger.debug("Took {} us to read {} records", watch.elapsed(TimeUnit.MICROSECONDS),
// c.getRecordCount());
spilledBatches--;
currentContainer.zeroVectors();
Iterator<VectorWrapper<?>> wrapperIterator = c.iterator();
for (VectorWrapper w : currentContainer) {
TransferPair pair =
wrapperIterator.next().getValueVector().makeTransferPair(w.getValueVector());
pair.transfer();
}
currentContainer.setRecordCount(c.getRecordCount());
c.zeroVectors();
return c;
}
public void addBatch(VectorContainer newContainer) throws IOException {
assert fs != null;
assert path != null;
if (outputStream == null) {
outputStream = fs.create(path);
}
int recordCount = newContainer.getRecordCount();
WritableBatch batch = WritableBatch.getBatchNoHVWrap(recordCount, newContainer, false);
VectorAccessibleSerializable outputBatch = new VectorAccessibleSerializable(batch, allocator);
Stopwatch watch = Stopwatch.createStarted();
outputBatch.writeToStream(outputStream);
newContainer.zeroVectors();
logger.debug(
"Took {} us to spill {} records", watch.elapsed(TimeUnit.MICROSECONDS), recordCount);
spilledBatches++;
}
