Re: Non-greedy wildcard possible? (Long)

[Hans Aberg]

At 04:31 +0200 2004/05/17, Magnus Lie Hetland wrote:
Please cc the Help-Bison list in sequel replies.

>> >I'm working on a project (atox.sf.net) where I want to use Bison to
>> >add markup to plain text. I need a rule to represent the plain text
>> >segments (between the markup tokens) and that's proving a challenge.
>>
>> Normally, this is done by the lexer, sending back say a token "text". Why
>> is you not doing it that way?
>
>Because what tokens are significant (as opposed to "plain text")
>is described by a user-specified context-free grammar.

Normally this is handled by a look-up table that is accessed by the lexer,
also in the case when the grammar is not formally context-free. By tweaking
the lexer and the parser together, one can often handle languages that are
not formally context-free.

So if you say you "already have solved the lexer part", that is a
suspicious remark, because the lexer will in practise depend on the
limitations of the parser.

I think you need to describe a little better your language problem.
Examples? Then, one can determine what techniques you might need.

  Hans Aberg