@Override
public StreamExpression toExpression(StreamFactory factory) throws IOException {
// function name
StreamExpression expression = new StreamExpression(factory.getFunctionName(this.getClass()));
// collection
expression.addParameter(collection);
// workers
expression.addParameter(
new StreamExpressionNamedParameter("workers", Integer.toString(workers)));
// stream
if (tupleStream instanceof Expressible) {
expression.addParameter(((Expressible) tupleStream).toExpression(factory));
} else {
throw new IOException(
"This ParallelStream contains a non-expressible TupleStream - it cannot be converted to an expression");
}
// sort
expression.addParameter(new StreamExpressionNamedParameter("sort", comp.toExpression(factory)));
// zkHost
expression.addParameter(new StreamExpressionNamedParameter("zkHost", zkHost));
return expression;
}
public ParallelStream(StreamExpression expression, StreamFactory factory) throws IOException {
// grab all parameters out
String collectionName = factory.getValueOperand(expression, 0);
StreamExpressionNamedParameter workersParam = factory.getNamedOperand(expression, "workers");
List<StreamExpression> streamExpressions =
factory.getExpressionOperandsRepresentingTypes(
expression, Expressible.class, TupleStream.class);
StreamExpressionNamedParameter sortExpression = factory.getNamedOperand(expression, "sort");
StreamExpressionNamedParameter zkHostExpression = factory.getNamedOperand(expression, "zkHost");
// validate expression contains only what we want.
if (expression.getParameters().size()
!= streamExpressions.size() + 3 + (null != zkHostExpression ? 1 : 0)) {
throw new IOException(
String.format(Locale.ROOT, "Invalid expression %s - unknown operands found", expression));
}
// Collection Name
if (null == collectionName) {
throw new IOException(
String.format(
Locale.ROOT,
"invalid expression %s - collectionName expected as first operand",
expression));
}
// Workers
if (null == workersParam
|| null == workersParam.getParameter()
|| !(workersParam.getParameter() instanceof StreamExpressionValue)) {
throw new IOException(
String.format(
Locale.ROOT,
"Invalid expression %s - expecting a single 'workersParam' parameter of type positive integer but didn't find one",
expression));
}
String workersStr = ((StreamExpressionValue) workersParam.getParameter()).getValue();
int workersInt = 0;
try {
workersInt = Integer.parseInt(workersStr);
if (workersInt <= 0) {
throw new IOException(
String.format(
Locale.ROOT,
"invalid expression %s - workers '%s' must be greater than 0.",
expression,
workersStr));
}
} catch (NumberFormatException e) {
throw new IOException(
String.format(
Locale.ROOT,
"invalid expression %s - workers '%s' is not a valid integer.",
expression,
workersStr));
}
// Stream
if (1 != streamExpressions.size()) {
throw new IOException(
String.format(
Locale.ROOT,
"Invalid expression %s - expecting a single stream but found %d",
expression,
streamExpressions.size()));
}
// Sort
if (null == sortExpression
|| !(sortExpression.getParameter() instanceof StreamExpressionValue)) {
throw new IOException(
String.format(
Locale.ROOT,
"Invalid expression %s - expecting single 'sort' parameter telling us how to join the parallel streams but didn't find one",
expression));
}
// zkHost, optional - if not provided then will look into factory list to get
String zkHost = null;
if (null == zkHostExpression) {
zkHost = factory.getCollectionZkHost(collectionName);
if (zkHost == null) {
zkHost = factory.getDefaultZkHost();
}
} else if (zkHostExpression.getParameter() instanceof StreamExpressionValue) {
zkHost = ((StreamExpressionValue) zkHostExpression.getParameter()).getValue();
}
if (null == zkHost) {
throw new IOException(
String.format(
Locale.ROOT,
"invalid expression %s - zkHost not found for collection '%s'",
expression,
collectionName));
}
// We've got all the required items
TupleStream stream = factory.constructStream(streamExpressions.get(0));
StreamComparator comp =
factory.constructComparator(
((StreamExpressionValue) sortExpression.getParameter()).getValue(),
FieldComparator.class);
streamFactory = factory;
init(zkHost, collectionName, stream, workersInt, comp);
}
