The simple answer is: when is it easier to write rules that describe your grammar than to write code that accepts the language described by your grammar.
If the only thing you need to parse looks the same as what you wrote above, then I would say that you could just write it by hand.
More generally, I would say that most common languages can be parsed faster manually (using a regular expression).
If you understand a context-free language with many rules and productions, ANTLRs (or other parser generators) can make life much easier.
Also, if you have a simple language that you expect to become more complex in the future, it will be easier to add rule descriptions to the ANTLR grammar than to create them in a hand-held encoder.
source share