[stringtemplate-interest] Iteration over multi-valued attribute using implied counter $i$ and first() rest()

[Terence Parr]

Hi. I believe all you need is:

make_keys_comparison(keys) ::= <<
$keys:{record[$i$] == $it$)}; separator=" and "$
 >>

Ter
On Jan 8, 2009, at 2:16 PM, Michael Albanese (malbanes) wrote:

> I have a need to traverse a multi-valued attribute utilizing the
> built-in iteration counter $i$ in addition to special casing for
> elements first() then rest().
>
> The following worker template that is passed a vector of strings  
> almost
> works for my needs.
>
> make_keys_comparison(keys) ::= <<
> $first(keys):{(record[$i$] == $it$)}$ $rest(keys):{ and (record[$i$]  
> ==
> $it$)}$
>>>
>
> I am using the implied $i$ iteration counter for part of the source
> generation as it conveniently could serve as the field offset for my
> generated lua source code.
>
> ...my desired output is an 'if' condition in lua source code. So in a
> good world I want the following output where keys = { agent slot  
> module
> }.
>
> if (record[1] == agent) and (record[2] == slot) and (record[3] ==
> module) then
>    return record
> end
>
> ...where the make_keys_comparison() template produces the conditionals
> inside the 'if' and the generated array offsets graduate sequentially
> from 1 to 'n'.
>
>
> What I get however is slightly off, and I believe has to do with the
> scoping of the implied $i$ counter. Apparently a new instance of the  
> $i$
> counter is scoped for both the first() and rest() attribute operators,
> and for my needs this hurts as evidenced by the actual output.
>
> if (record[1] == agent) and (record[1] == slot) and (record[2] ==
> module) then
>    return record
> end
>
> Here one can see that the record[] offsets are not accurate. They are
> affected apparently by the scoping of $i$ between first() and rest().
>
> I'd like to know if using the $i$ is the wrong tact here (or perhaps
> there is a bug and this should work ?).
>
> If not is there a way to count and display that count value along with
> each element(s) in the multi-valued 'keys' attribute while also
> retaining the special casing needed for first/rest ?
>
> Thanks,
> Michael
>
>
>
>
>
>
>
>
> _______________________________________________
> stringtemplate-interest mailing list
> stringtemplate-interest at antlr.org
> http://www.antlr.org/mailman/listinfo/stringtemplate-interest