http://www.antlr.org:8080/pipermail/antlr-interest/2005-September/013603.html

Let me know, if you have any pointers to this problem.

Thanks,
manju

Antlr does support the integration of multiple grammars and even grammar extension through the notion of “vocabularies”. You might check out the antlr doc on Vocabularies for some more info.

Firstly, I was wondering if you could discuss how to properly handle errors and report them(line no., module, etc). instead of the cryptic ‘exception thrown’ kind of errors. I’ve seen a section on error recovery in the antlr’s documentation, but it’s such a mind-bender :-) I’m still new at all this, so pl. excuse my ignorance.

Secondly, what’s the best way to parse included modules, e.g. if a script includes another script(a la C includes), I would like to parse
the included file first. Any links to articles on this subject would be great too.

Thanks again
Manju
-manju

Thanks for pointing this out! As it turns out, I forgot something critical in the parser definition.

But first, the reason you’re seeing an error with characters other than those in the punctuation section is that these are not valid according to the lexer, so the lexer is throwing an error.

In the case of the extra semicolon, the lexer does understand the token, so no error is thrown (by the lexer). If you look at the output of the TestParser program with your example script, you’ll notice that it prints only the first block, which was my first big clue to my mistake. The key is that I didn’t tell Antlr to parse the *whole* input - so instead Antlr just consumed input and matched it as asked until it was no longer valid, in which case it just stopped, leaving a slew of tokens unparsed.

Obviously, this is not what we want. To fix this, we need to add the pre-defined EOF token to the end of our script rule - this tells Antlr that we expect to match the entire input stream and end with EOF:

script : (block)* EOF

Then if we run with your script we will see the expected error:

Parsing: test/puredanger/parser/comment.script
line 7:5: expecting EOF, found ‘;’

Thanks again and I’ll update the article appropriately.