docx4j is working very well for us. Thanks!
When updating a docx file, with a repeating row, multiple times via preprocess/applybindings, the XPathIDs and XPaths of the cell contents have the "row"/etc. numbers expanded, which get longer and longer with each update via XML.
Before I dig further into it, is there a simple (or not so simple) method to, prior to the next XML update,
1. delete all the table rows except the first ...