Processing Streaming Data - Real-Time Analytics - page 175

The most basic local operation is the Filter , which is implemented by
extending the BaseFilter class. A single method, isKeep , is
implemented todecidewhetherornotaTuple'sprocessing shouldcontinue.
For example, this filter implementation randomly chooses to keep half of its
results:
public class RandomFilter extends BaseFilter {
private static final long serialVersionUID = 1L;
Random rng = new Random();
public boolean isKeep(TridentTuple tuple) {
return rng.nextFloat() > 0.5;
}
}
The next simplest operation is the Function , which is implemented by
extendingthe BaseFunction class(someTridenttutorialsincorrectlyhave
filters extend BaseFunction as well, but this is not the case in 0.9.0).
In functionality, Function objects are closest to the BasicBolt in the
“traditional” Storm topology. They take in a single Tuple and can produce
zero or more tuples, which are appended to the original tuple. If no tuples
are produced, the original tuple is filtered from the stream. For example,
this Function implementation adds a random number to each incoming
Tuple :
public class RandomFunction extends BaseFunction {
private static final long serialVersionUID = 1L;
Random rng = new Random();
public void execute(TridentTuple tuple,
TridentCollector collector) {
collector.emit( new Values(rng.nextDouble()));
}
}
Both Filter and Function classes use the each method to attach
themselves to a stream. For filters, the each method takes the form of
each(inputFields,filter) :