CEDET

On 09/26/2012 09:07 AM, kototama kototama wrote:
> Hi,
>
>
> I'm trying to parse this statement:
> (def () x 42)
>
> with the following grammar:
> https://github.com/kototama/clojure-semantic/blob/78f00ca4369a3df0b2fbb81ba5b72e1f90b3008c/clojure.wy
>
> The
> DEF list SYMBOL clojure-expr
> rule is matched but when I call M-x semantic-complete-jump the symbol
> does not appear.
>
> If I restrict the definition to DEF SYMBOL clojure-expr
> and parses (def x 42) then it works (building with $2 instead of $3 in
> this case).
>
> I also tried DEF LPAREN RPAREN SYMBOL clojure-expr  and it didn't work.
>
> What's the problem with the empty list then? May it be related to the
> way blocks work?

Hi,

The semantic parsers treat lists such as () or {} special.  Any 
start/end characters can be turned into a list with the <block> token 
type as you used it.

That means that

(def () x 42)

should be converted lexically into:

DEF PAREN_BLOCK SYMBOL

which you could verify with semantic-lex-debug in your clojure buffer.

Once you match the PAREN_BLOCK, then you use EXPANDALL to look inside. 
For example, your rule:


list: LPAREN RPAREN
     | LPAREN coll-content RPAREN
       (EXPANDFULL $2 coll-content)
     ;

should instead be:

list: PAREN_BLOCK
       (EXPAND $1 coll-content)
     ;

and your coll-content in this case is some arg-list parser instead.

Does that make sense?

Eric

> (def () x 42)
>
> should be converted lexically into:
>
> DEF PAREN_BLOCK SYMBOL

There must be a problem in the lexer definition or usage because even
if I used the
wisent-clojure-wy--<block>-block-analyzer

I get LPAREN and RPAREN with semantic-lex-debug instead of getting PAREN_BLOCK.

What could be the problem?

On 10/05/2012 10:48 AM, kototama kototama wrote:
>> (def () x 42)
>>
>> should be converted lexically into:
>>
>> DEF PAREN_BLOCK SYMBOL
>
> There must be a problem in the lexer definition or usage because even
> if I used the
> wisent-clojure-wy--<block>-block-analyzer
>
> I get LPAREN and RPAREN with semantic-lex-debug instead of getting PAREN_BLOCK.

Is the variable `semantic-lex-depth' set to something larger than 0?  I 
looked at your code, but I didn't see a setting for it.

That depth variable will cause the lexer to dig into lists instead of 
creating blocks.

Alternately, if the lexer for parens shows up in front of the  BLOCK 
analyzer, it will do that first, and skip the blocks.  Your regex 
analyzer in particular looks like it might match parens.  Unless you 
need it right off, I'd comment that one out and see how it goes.

Your best bet is to build up code examples, and use semantic-lex-debug, 
and make sure you are getting the right lexical token stream before 
worrying much more about the rules in your grammar.

Eric

> Is the variable `semantic-lex-depth' set to something larger than 0?  I
> looked at your code, but I didn't see a setting for it.

The variable is set to the zero.

So I tried a very minimal example, that just parses such BLOCKs:

https://github.com/kototama/clojure-semantic/tree/trashall

When trying semantic-lex-debug on a buffer containing only "()" I
still only get LPAREN and RPAREN token.


Maybe something is wrong in the way I setup the parser?