[MOARDONGZPLZ]

Looks like it’s just scrambling each individual word. Seems straightforward to programmatically look for groups of things that aren’t legitimate words on a page.