[antlr-interest] ANTLR Questions

[Thomas Brandon]

On Wed, Jan 21, 2009 at 11:55 PM, Bruno Marc-Aurele
<bruno.marc-aurele at polymtl.ca> wrote:
> Thanks a lot Johannes.
>
> Has anyone used heterogenous trees? I would like to know how customisable my AST
> can be... I will rewrite my question concerning this so you don't need to look
> for it in the previous messages:
>
> Q: I read the section about heterogenous nodes (using the <X> syntax), but all
> the examples seem applied on token calls (ex: ID<V> and 'id'<V>). Does it work
> on rule calls (ex: rule<V>)?
ANTLR creates Abstract Syntax Trees not parse trees. So nodes are not
automatically created for rules. You have to manually add such
imaginary nodes specifying a token type, so the rule<V> syntax does
not make sense. You would need to do something like:
rule: children -> ^(RULE<V> children);

> Another question pops for me: can I define new attributes for a node in semantic
> actions? For example, if I write the following rule:
>
> rule: ID subrule {$subrule.newAttribute = 1;} INT;
>
> ... will my node object have a "newAttribute" property (C# target) when
> generated?
ANTLR doesn't generate the custom AST types for you, you must manually
create them.

Tom.
>
> Thanks a lot,
> Bruno M-A.
> --
> Bruno Marc-Aurèle
> courriel Poly: bruno.marc-aurele at polymtl.ca
> courriel perso: bruno.m-a at hotmail.com
>
>
> Selon Johannes Luber <jaluber at gmx.de>:
>
>> Bruno Marc-Aurele schrieb:
>> > Hello everyone,
>> >
>> > I am currently working on a project where I need to translate files (mainly
>> > code) from one technology to another. I already have a LL(*) grammar for
>> the
>> > source language and I am considering rewriting this grammar using ANTLR to
>> > benefit from C# generation target.
>> >
>> > I have used JavaCC a lot in the past and I have some questions about ANTLR,
>> a
>> > tool that I find really promising for my project. I read the documentation,
>> but
>> > some things just still are unclear.
>> >
>> > 1) If I want to generate a parse tree, do I HAVE to use a tree grammar or
>> will a
>> > combined one work?
>>
>> You create an AST via a parser grammar. A tree grammar parses only
>> created ASTs and may rewrite it, too.
>> >
>> > 2) About the generated tree, what will be the types of the nodes? Will each
>> rule
>> > have its type (ex, rule test would have a Test() type node) or are all the
>> nodes
>> > the same type (NodeType, for example)?
>>
>> Usually, the used class is BaseTree, but can be changed via a grammar
>> option. Heterogenous are also supported, but I have never used them, so
>> I can't answer the last question.
>>
>> Johannes
>> >
>> > 2.1) I read the section about heterogenous nodes (using the <X> syntax),
>> but all
>> > the examples seem applied on token calls (ex: ID<V> and 'id'<V>). Does it
>> work
>> > on rule calls (ex: rule<V>)?
>> >
>> > Your answers might spark new questions from me, but for now it's all I need
>> to
>> > know.
>> >
>> > Thanks a lot for your support,
>> > Bruno Marc-Aurèle.
>> >
>> >
>> >
>> >
>> >
>> > List: http://www.antlr.org/mailman/listinfo/antlr-interest
>> > Unsubscribe:
>> http://www.antlr.org/mailman/options/antlr-interest/your-email-address
>> >
>>
>>
>
>
> List: http://www.antlr.org/mailman/listinfo/antlr-interest
> Unsubscribe: http://www.antlr.org/mailman/options/antlr-interest/your-email-address
>