The core transforms of the DITA Toolkit represent an implementation of all processing defined by OASIS in the DITA specification.
A pre-process is done before the main transformation. The input of the pre-process is dita files (maps and topics) and the output of the pre-process is often referred to as "normalized" dita files. The pre-process stage resolves several common DITA behaviors, such as resolving conref attributes, resolving keyref values, and adding links based on the hierarchy and relationship tables in a map. The normalized dita files are in a temporary directory. Most DITA transforms should use this common pre-process setup.
The complete transformation including pre-process can be executed by an ant script. There are some examples of simple ant scripts in the directory samples/ant_sample. See Running Ant for more details about Ant processing.