Processing Streaming Data - Real-Time Analytics - page 158

A Filter and Split Bolt
This example shows a bolt that takes an input stream and splits it into
several different streams according to a regular expression on a field. It
also shows how to introspect the incoming tuples to define the output
streams programmatically.
The filters are defined during the configuration of the topology. The
bolt defines a simple internal class called FilterDefinition , which
holds the output stream, the name of the field to check, as well as the
regular expression that will be evaluated against the field's value. This
class implements Serializable so it will be properly serialized when
the bolts are distributed across the cluster.
The filter function itself is implemented in a “chainable” style to
make it easy to use in topology definitions:
public class FilterBolt extends BaseRichBolt {
private static final long serialVersionUID =
-7739856267277627178L;
public class FilterDefinition implements
Serializable {
public String stream;
public String fieldName;
public Pattern regexp;
public Fields fields;
private static final long serialVersionUID = 1L;
}
ArrayList<FilterDefinition> filters
= new ArrayList<FilterDefinition>();
public FilterBolt filter(String stream,String
field,String regexp,
String... fields) {
FilterDefinition def = new FilterDefinition();
def.stream = stream;
def.fieldName = field;
def.regexp = Pattern. compile (regexp);