I am afraid we will need more context than this to determine the root-cause. Are you able to provide us with the details of where in the Hibernate code is the call ingot StringHelper#replace at least happening?

While certain input values could make #replace problematic as a corner case bug, I am not seeing any recent changes to the helper class that stand out to me as a reason why you would observe differing behavior between 5.2 and 5.4; at least based on the helper class logic. Perhaps a call reference was changed/added. Knowing the initially call point where you observe this could at least give us some clue.