As I said, a lot depends on what you're trying to do, but I will throw
out some more suggestions, this time dealing with events.
There's a full set of events, of which the most interesting to you may
be "pause before lexeme":
https://metacpan.org/pod/Marpa::R2::Scanless::DSL#pause
Also, in defining a lowest priority catch-all, which I take it is the
intent with
Error ~ [\s\S]*
you might be more successful if you make it shorter, perhaps like
Error ~ [\s\S]
The idea is that even LATM favors longest tokens in some circumstances,
so if you want a token to have truly the "match of last resort", it's
helpful to make it as short as possible. Here, you just try to match
the first character which, since there are no zero-length lexemes, is
sure to at least tie for shortest. Any longer ,match will cause
possible matches of <Error> to be ignored, which I am guessing is the
desired behavior. As for what the pattern specified for <Error>
actually matches, in the case of a "pause before" it does not matter, as
the lexeme will not actually be read.
Hope this helps, jeffrey
On 04/29/2014 07:47 AM, toloaca...@gmail.com wrote:
Hello,
I can extract from some data structures a set of grammar rules that
looks like:
topLevelRule1 ::= rule1 rule2 rule3
topLevelRule2 ::= rule2
rule1 ::= rule11 rule12
rule11 ~ token11
rule12~ token12
rule2 ~ token2
rule3 ~ token3
.... and so on (any kind of nested structures of this kind are to be
expected),
and I would like to create a Marpa grammar from this to
parse against some MathML stuff ( and replace subtrees according to it.)
Now the input is not completely described by the grammar, so I would
want to
use some kind of structure like:
Script ::= Notation (top level rule -> that is what I am going to
replace)
|| Presentation (other MathML not described by my grammar)
|| Content (stuff that was already replaced on a previous run or
by any other means)
|| Error (anything)
The problem is that the top level rules have precedence that has to be
taken into
account.
-> I have tried using || to implement that, but found out that its action
is not deterministic and hence useless for my purpose.
-> I have also tried to use ranking_method in order to implement this,
but it just doesn't work out.
So, my question is as follows - if I have a (potentially large) set of
alternative rules with
certain precedence, how do I define a grammar like that ( also taking
into
account the possibility of ambiguous parses) - so that I can use the
"event" adverb
to actually perform some substitutions.
To make it clearer - I would be happy enough even if I only got an
answer to how the following
should be implemented so that the alternative "|| Other action =>
do_printErr" is ONLY
executed if the first alternative failed.
:start ::= Script
Script ::= Number ',' Number action => do_printNr
|| Other action => do_printErr
Other ::= Error
Error ~ [\s\S]*
Number ~ digits
digits ~ [\d]+
Thank you in advance for help,
Toloaca Ion
--
You received this message because you are subscribed to the Google
Groups "marpa parser" group.
To unsubscribe from this group and stop receiving emails from it, send
an email to marpa-parser+unsubscr...@googlegroups.com
<mailto:marpa-parser+unsubscr...@googlegroups.com>.
For more options, visit https://groups.google.com/d/optout.
--
You received this message because you are subscribed to the Google Groups "marpa
parser" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to marpa-parser+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
