Apache Camel - Aggregate different object types - apache-camel

Question

I have a route that, when lauched, requests messages from two datasources (routeA and routeB) and aggregates them into a single message. Each aggregated message MUST contain exactly one routeA message and one routeB message, if not, then drop it.
This process must be launched at specific intervals (i.e: every 5 min).
My question is, how can I let the aggregator know that all messages from routeA and routeB where processed and the messages that didn't find their pair, must be droped?
I'm currently using completionTimeout feature, but I don't like this solution for obvious reasons.
I know camel has a completionFromBatchConsumer feature, but I don't know how to use it with with multiple datasets.
Am greatefull for any advice.
Here's what I have right now:
<!-- main route -->
<route id="main">
<camel:from uri="timer://timer1?period=20000"/>
<multicast>
<to uri="direct:startA"/>
<to uri="direct:startB"/>
</multicast>
</route>
<!-- messages from route A -->
<route id="routeA" />
<from uri="direct:startA" />
<to uri="sql:select * from sampleDB?dataSource=ds"/>
<split>
<simple>${body}<simple>
<marshal ref="ObjectAJsonConverter"/>
<unmarshal ref="ObjectAJsonConverter"/>
<to uri="bean:myProcessor?method=addObjectACorrelationKey"/>
<to uri="seda:myAggregator"/>
</split>
<!-- messages from route B -->
<route id="routeB" />
<from uri="direct:startB"/>
<to uri="ldap:ldapcontext?base=DC=company,DC=net"/>
<split>
<simple>${body}<simple>
<marshal ref="ObjectBJsonConverter"/>
<unmarshal ref="ObjectBJsonConverter"/>
<to uri="bean:myProcessor?method=addObjectBCorrelationKey"/>
<to uri="seda:myAggregator"/>
</split>
<!-- aggregate the messages, create new ObjectC that contains ObjectA and ObjectB -->
<!-- wait 200000 ms for all messages from routeA and routeB to enter the aggregator -->
<route id="aggretatorRoute">
<from uri="seda:myAggregator"/>
<aggregate ref="myEntityAggregator" completionSize="2" completionTimeout="200000" discartOnCompletionTimeout="true" ignoreInvalidCorrelationKeys="true">
<correlationExpression><simple>${in.header.objectid}</simple></correlationExpression>
<to uri="bean:myProcessor?method=doSomethingWithObjectC"/>
</aggregate>

Answer 1

You can just in your AggregationStrategy only aggregate one of ObjectA and one of ObjectB. So if you see a 2nd of either of them, then just not aggregate it. And if you then want to drop what you have done so far, then you can mark the exchange to stop, by setting
exchange.setProperty("CamelRouteStop", true);
And if you then want to drop this immediately, then add a completionPredicate, that checks if that stop has been set.
<completionPredicate><simple>${property.CamelRouteStop} == true</simple></completionPredicate>
And for the correlationExpression, you can likely just use <constant>true</constant> as it seems you only work on one group.

Answer 2

Thanks Claus. Actually I took a different approach. Instead using an aggregator to join three different object, I now query a list of ids and use those to progressively build my complex object.
<route id="composeObject">
<from uri="sql:select id from people?oneSource">
<split><simple>${body}</simple>
<to uri="direct:getobjectOne"/>;
<to uri="bean:addToComplexObject"/>
<to uri="direct:getObjectTwo/>
<to uri="bean:addToComplexObject"/>
<to uri="direct:getobjectThree/>
<to uri="bean:addToComplexObject"/>
<to uri="seda:outChannel"/>
</split>
</route>