As we have already seen, the parser (phase 2) is responsible for filling the source structure with data to create the source tree.

With this, you have generated the basic structure of the data on which the mapping in phase 3 is based. The data is divided into records in the source tree and in each record you have a hierarchical structure of nodes and fields. This is always the first critical design step when you think about what your output data should look like in the end.

The mapper's task is to build a destination tree from the source tree.

Starting from the source tree, you have two major design tools to achieve that - records and paths.