I tried to solve it two ways, once using a text input with the different "end-of-street-address" possibilities used to mark the transition from the street address to the city, and then second using a single RegEx Parse tool (including this logic within the parsing). I would say that although second solution looks nice because it uses a single parse tool to complete, potentially it would be nicer to maintain a single text input list if any new "end-of-street-address" words/characters were found when used on a real or different dataset.