[antlr-interest] Re: Syntactic predicates question

[Artem Dmytrenko]

Hmm, I'm really confused by the behavior then. "A12345" definitely doesn't 
match rule 'A' so (1) should fail and not consume the first character of 
the string. Shouldn't ANTLR examine at least k characters (in my case 
k=2, so it should be looking at 'A' and '1') from input stream before 
making a decision about which token matched? The generated code for 
matching 'A' in lexer is as follows:

if ((LA(1)=='A') && (true)) {
   match('A');
}

Shouldn't it be something similar to the following?

if ((LA(1)=='A') && (LA(2)==END_OF_TOKEN) {
   match('A');
}

I'm trying to use syntactic predicates for parsing a language with 
keywords that may be part of identifiers (e.g. keyword "Action", 
identifier "Action/*/123"). Is there a better approach than syntactic 
predicates to attack this scenario?

Thank you again for your help.

Sincerely,
Artem Dmytrenko

On Mon, 30 Jan 2006, Xue Yong Zhi wrote:

>
>
> Artem Dmytrenko wrote:
>
>> 
>> line 1:94: expecting ID, found 'A'
>> 
>> It appears that the match is stuck in the middle - e.g. ActionToken rule 
>> rejected the string but ID did not match it. Is that the expected behavior 
>> for syntactic predicates? Are there any workarounds for this problem?
>> 
>
> Your parser is thinking this way when parsing "A12345":
>
> 1. Try ActionToken, and match the first 'A'.
> 2. Try ActionToken again with the rest of the input "123456", do not match.
> 3. Then try ID, still no match.
> 4. Give you the warning.
>
> Most of the time Antlr does not follow "the longest one that matches wins" 
> rules.
>
> -- 
> Xue Yong Zhi
> http://seclib.blogspot.com
>
>