[antlr-interest] lexer state and SMILES strings

[Andrew Dalke]

On Nov 7, 2007, at 7:05 PM, Austin Hastings wrote:
> I'm not loving the language, but I guess you can't change it. :(

No I can't.  Though I'm rather fond of it.  It's a very
malleable notation and has been around for about 20 years.
I can definitely point to worse ;)


> My first recommendation would be to use antlr2 -- antlr3 wants to  
> be able to support all that stuff, but can't at the moment.

Ahh, didn't know that.  Figured I'ld go for the latest and greatest.

> Next: go ahead and use the parser as a lexer,
> and then use a tree parser to handle the output.
> Don't get hung up on the names, just work with what works.

> For parsing the insides of square brackets, have a look at
> the book's section on emitting multiple tokens.

Ahh, hmm.  Yeah, that part of the language is simple and I can
drop in code from other lexers, like re2c, in that section.
I think.  Error reporting/handling won't be nice, but I don't
need that for now.

Thanks!

				Andrew
				dalke at dalkescientific.com