[antlr-interest] How to communicate between rules in a tree grammar

[Jim Idle]

Though you specify x and y as return, there is also still a $st. You can
either build the $dst in the b rule or use the results of the b rule in a
caller to build a template, but b won't build you anything by default
because you changed the return pattern. So use the bs return, create a
template using x and u, add that template to a list of template, pass that
list to the template constructor and not the raw $bs.

Jim

> -----Original Message-----
> From: antlr-interest-bounces at antlr.org [mailto:antlr-interest-
> bounces at antlr.org] On Behalf Of Sanchay Harneja
> Sent: Wednesday, April 20, 2011 5:48 AM
> To: antlr-interest at antlr.org
> Subject: [antlr-interest] How to communicate between rules in a tree
> grammar
>
> Hi,
>
> I am new to ANTLR, using it to create a DSL. So far I've been able to
> write a parser grammar, a tree grammar & string template file. Its all
> working as I expect.
>
> But I have a niggling feeling that I may not be using the tree grammar
> & stg in an idiomatic manner. I'll ask my question using an example,
> suppose the tree grammar has rules a, b, c:
>
> options {
>     ...
>     output=template;
> }
>
> a :
>     // ...
>     -> template( .. )
>
> b : returns [ int x, int y ]
>     // only return no string template
>
> c : b (as+=a)+
>     -> c(x={$b.x}, y={$b.y}, as={$as})
>
> This works fine, but is it an "correct" way? In the generated code the
> type of $b is b_return while the type of $as is List(containing
> StringTemplate objects).
>
> Now suppose I have a rule like:
>
> d : a (bs+=b)+
>      -> d(a={$a}, bs={$bs})
>
> This doesn't work, as the type of $bs is List containing StringTemplate
> objects, but not b_return objects. Since I haven't defined a
> StringTemplate for b, this doesn't work.
>
> Maybe this is the right thing to do, i.e., in rule c type of $b is
> b_return, whereas in rule d type of $bs is List(containing
> StringTemplates as opposed to b_returns).  But then how I communicate
> the structure of b (i.e. x, y) to d ? One way is maintaining a global
> memory for this. Is there any other way?
> How would you guys tackle this issue.
>
> Thanks!
> Sanchay
>
> List: http://www.antlr.org/mailman/listinfo/antlr-interest
> Unsubscribe: http://www.antlr.org/mailman/options/antlr-interest/your-
> email-address