row-accumulate

(See http://www.therealvan.com/rowaccumulator.html for write-up page.)

Accumulator class to help partition and deserialize an ordered data stream (e.g. CSV lines, JDBC ResultSets) into data beans, supporting chaining to build up collections of subpartitions.

Given a data stream of this format (example in CSV):

ID, firstName, lastName, dish, identity_type, identity_value
100, Joe, Slim, Chicken, Driver's License, B1234567
100, Joe, Slim, Chicken, SSN, 123-45-678
101, Bob, Portly, Fish, Driver's License, C1234567
101, Bob, Portly, Spinach, Driver's License, C1234567
101, Bob, Portly, Steak, Driver's License, C1234567

Three beans are created to model the top-level Person entity, then the children (siblings of one another) entities for Identity and Dish. It is important that the data streams are ordered to group key properties in contiguous rows of data. (This corresponds to the use of ORDER BY clauses in SQL.) For example the above may be achieved by executing an SQL like:

select ID,firstName,lastName,dish,identity_type,identity_value
  from ...
  order by ID, dish, identity_type;

Correct use of the Accumulator class can produce a collection of Person instances those identities and dishes collections contain the extracted entities. E.g.:

        people list: [
          {100:Slim, Joe;
            Identities: [ {Driver's License:B1234567} {SSN:123-45-678} ];
            Dishes: [ {name:Chicken} ]},
          {101:Portly, Bob;
            Identities: [ {Driver's License:C1234567} ];
            Dishes: [ {name:Fish} {name:Spinach} {name:Steak} ]}
        ]

The small class Accumulator makes heavy use of generics and closures in order to offer flexibility and reusability to address different input sources, key calculation logic, and bean implementations. See the example domain classes under the org.van.rowaccumulate.example package, starting with the unit test class org.van.rowaccumulate.AccumulatorTest, for how to use it.