get some data from the database into esper

test.java

import com.espertech.esper.client.*;

public class test {
    
    public static void main(String[] args) throws InterruptedException {
        
    	Configuration configuration = new Configuration();		
    	configuration.configure();
    	
        EPServiceProvider svcProvider = EPServiceProviderManager.getDefaultProvider(configuration);
        EPRuntime runtime = svcProvider.getEPRuntime();
        
        EPAdministrator epAdmin = svcProvider.getEPAdministrator();
        
        // This will trigger a SensorEvent for every result from the MySQL query, for every TriggerQryEvent
        //   The ORDER BY is critical, because that determines the order in which the SensorEvents are
        //   posted.
        epAdmin.createEPL("create schema TriggerQryEvent as (triggerId int)");
        epAdmin.createEPL(
    		"insert into SensorEvent " +
    		"select sqldata.* from " +
        		"sql:mydb['SELECT " +
        		"    UNIX_TIMESTAMP(Record_Time)*1000 + MICROSECOND(Record_Time)/1000 as RecordTime, " +
        		"    d.Sensor_ID, Sensor_Name, Value " +
        		"FROM fdr_FDRData AS d join fdr_SensorLU AS s on d.Sensor_ID = s.Sensor_ID " +
        		"WHERE aircraft_ID = 18 "+
        		"ORDER BY Record_Time'] sqldata, " + 
        		"TriggerQryEvent ");
        
        // Create MachEvents for those SensorEvents that match the Sensor_Name we want ..
        //  Being able to do this efficiently is a key difference between using Esper &
        //  doing straight-up database queries, which we could easily do for this simple
        //  example, but would fall apart as we add more and more complex event queries.
        epAdmin.createEPL(
    		"insert into MachEvent " +
    		"select * from SensorEvent(Sensor_Name = 'Mach')");
        
        // Find all 14 second windows having min(Value) >= .7.  With this query, 2 consecutive
        //  windows that satisify the condition will return 2 separate events.
        // ****** We don't do anything w/ the output of this query; it's just an example ******
        epAdmin.createEPL(
        	"insert into OverSpeedEvent " +
    		"select min(Value), max(Value), min(RecordTime), max(RecordTime) as mt " + 
        		"from MachEvent.win:ext_timed(RecordTime, 14 sec) " +
    		"having min(Value) >= .7");

        // The above query releases a flurry of events, when really there's only 1 long one.
        //   This one will detect the transition; still not really what we want, but closer:
        // ****** We don't do anything w/ the output of this query; it's just an example ******
        epAdmin.createEPL(
        	"select * from pattern [ every (ee=MachEvent(Value >= .7) -> MachEvent(Value < .7)) ]");
        
        // This one finally does what we want.
        EPStatement epl = epAdmin.createEPL(
        	"select startTime, startVal, endTime, endVal, endTime-startTime as durationMillis " +
        	"from MachEvent match_recognize ( " +
        			"measures S[0].RecordTime as startTime, S[0].Value as startVal," +
        			"         E.RecordTime as endTime, E.Value as endVal " +
        			"pattern ( S+? E ) " +
        			"define S as S.Value >= .77, E as not E.Value >= .77 " + 
        	") having endTime-startTime > 5000");
        
        
        
        // This adds something to the last statement to handle the detected events ..
        //  all we do is print them to stdout.
        epl.addListener(new UpdateListener() {
			@Override
			public void update(EventBean[] newEvents, EventBean[] oldEvents) {
	        	System.out.println(newEvents[0].getUnderlying());
			}
		});
        
        // This triggers the TriggerQryEvent event, which will cause the 
        //   SQL query to be executed and each row to by posted as a SensorEvent.
        //   There's probably a way to do this w/ pure EPL, but I didn't find it quickly.
        EventSender sender = runtime.getEventSender("TriggerQryEvent");
        sender.sendEvent(new java.util.HashMap<String, Integer>());
    }
}