The XRCS (Xerces) api's will allow you to start either a DOM or SAX engine and then navigate the XML (DOM) or consume tags as they are parsed (SAX).
See the following guide: JD Edwards EnterpriseOne Tools 8.98 Development Tools: APIs and Business Functions Guide - "Using the SAX Parser"
For examples of how these APIs are used you can search your E812\PathCode\Source folder for C source files containing "XRCS". B01RS056.c is a good example of how to use the DOM parser. The function is part of RapidStart and parses the XML specs for UBE data selection. B34A2270 is a good example of the SAX engine. It processes an inbound XML forecast file.
I have used both engines with success. DOM is easier to navigate the XML with but has a higher memory footprint as the entire document is loaded into memory. SAX is a stream parser and has a small memory footprint. The downside is that you must process the XML document tag-by-tag as it is fed to you. You cannot query or navigate the XML document with a SAX parser.
For XML files that will be imported into JDE tables I would suggest the SAX parser. It will allow you to process files of almost any size without worrying about memory footprint.