Differentiating between several parser and lexer issues

[ftomassetti]

I think that, in order to facilitate the translation of error messages, we need to produce different codes for different messages. We also need to provide all the arguments needed to recreate the messages in another language.

Unfortunately, that means looking at the actual strings being generated unless we replace the DefaultErrorStrategy with a different one (if you prefer that I go that route, let me know, and I will update the PR).

Note also that in some cases, we produce more than one argument with the same name, implicitly providing an array. This is the case, for example, for cases where we need to report multiple tokens that could be used at a certain point. We could report all of them in a single string with commas, but perhaps in certain languages, they do not use commas

Fix #71

[ftomassetti]

Also, probably we will need to distinguish across more error types in the future, judging by the different strings produced in: https://github.com/antlr/antlr4/blob/7d4cea92bc3f7d709f09c3f1ac77c5bbc71a6749/runtime/Java/src/org/antlr/v4/runtime/DefaultErrorStrategy.java

[alessiostalla]

While a custom error strategy would make this cleaner, given that the users of the library are expected to instantiate the ANTLR parser themselves, I don't think it's a great idea to impose our own custom strategy. We could try opening an issue on antlr4ng and/or the ANTLR main project.

[ftomassetti]

Thank you @alessiostalla ! When you have time, could you do a new release?

[alessiostalla]

Published now!

[ftomassetti]

Thanks!