Dozer is a fast and flexible framework for mapping back and forth between Java Beans. Coupled with Camel's automatic type conversion, it's a formidable tool for dealing object to object mapping headaches that crop up in enterprise integration projects.
To explain how Dozer can be uses within Camel we'll use the following example of a simple Customer Support Service. The initial version of the Service defined a 'Customer' object used with a very flat structure.
In the next version it was decided to structure the data better in the model by moving the address data into its own type, with the resultin domain object ending up looking like
Much nicer! But as often occurs, the previous version of the service, with the old flat 'Customer' object, was in production with a client and the project must support the legacy interface. To support both versions, we must add a mechanism to convert between the old Customer service type and the new Customer domain type and back again. It would be a simple matter to write a custom converter class to map between them, but this may not be the only service/domain inconsistency and these tedious and error prone custom mappings could quickly start to add up, and bugs with them.
To a large extent the two object share identical structure, with only the address representation being different. It would be very helpful if there were a practical way to to automate this kind of mapping, such that the similar properties could get mapped automatically and only the inconsistencies requiring custom mapping.
This is where Dozer comes in; It uses reflection to map data between two bean types using a set of simple mapping rules. Where no rule is specified, dozer will attempt to map between them by using matching properties of two beans. In this way focus can be given to the inconsistencies between the beans i.e. the address properties, knowing that dozer will automatically match and convert the others.
Dozer's configuration is extremely flexible and many mapping scenarios are covered here. For our simple example, the configuration looks like the following.
Camel provides a simple mechanism to integrate Dozer Mappers with it's own powerful Type Conversion framework. Its configured by creating an instance of DozerTypeConverterLoader providing it the camel context and an optional Dozer mapper. If no mapper is supplied, Camel's registry will be searched for suitable instances. The loader will query the Dozer Mapper for the the types it converts and a register them with Camel's type conversion framework to be handled by the mapper.
In Java it can be configured as follows:
Or in Spring
Now, where necessary, Camel will use Dozer to do conversions; In our case between the new domain and legacy Customer types e.g.