[antlr-interest] nondeterminism issue for dummies ?

[Loring Craymer]

ANTLR can't see past the leading whitespace; you want to do something like

 

(           ( (WS)* "axi.") => (WS)* "axi."                  // include
whitespace in synpred

            (           "locate"

                        |

                                    "reject"

            )

            |

                        (~('a')) (~('\n'|'\r'))*

)

CRLF

 

The non-determinism is real in this case.

 

--Loring

 

  _____  

From: antlr-interest-bounces at antlr.org
[mailto:antlr-interest-bounces at antlr.org] On Behalf Of Michiel Vermandel
Sent: Monday, July 10, 2006 12:59 AM
To: antlr-interest at antlr.org
Subject: [antlr-interest] nondeterminism issue for dummies ?

 


Hi, 

another nondeterminism issue... 

The issue is in deciding between a single-line comment rule or a directive
rule. 

A single-line comment is "--" (two minus signs) followed by anything untill
the end of the line. 

A directive is "--"(two minus signs) follwed by spaces or tabs, then
followed by some reserved words such as "axi.locate" or "axi.reject" 

So: 

"-- hello world" CRLF 

is a comment 

"-- axi.locate blabla" CRLF 

is a directive. 

My rule works though gives a nondeterminism warning. 

Rule: 
-------------------------------- 

SL_COMMENT 
       :        "--"!  ( 
              ( 
                 (' '|        '\t')* 
                 (   
                    ("axi.locate") => (  "axi.locate" { _ttype =
AXI_DIRECTIVE; } ) 
                   |("axi.reject") => (  "axi.reject" { _ttype =
AXI_DIRECTIVE; } ) 
                 ) 
               ) 
               | (~('a')) (~('\n'|'\r'))* 
           ) 
           CRLF 
       ; 

Warning: 
------------------------------ 
ANTLR Parser Generator   Version 2.7.5rc2 (2005-01-08)   1989-2005 jGuru.com

FormsLexerSnip.g:20:24: warning:lexical nondeterminism between alts 1 and 2
of block upon 
FormsLexerSnip.g:20:24:     k==1:'\t',' ' 
FormsLexerSnip.g:20:24:     k==2:'\t',' ','a','x' 
FormsLexerSnip.g:20:24:     k==3:'\t',' ','a','i','x' 
FormsLexerSnip.g:20:24:     k==4:'\t',' ','.','a','i','x' 


Any ideas how to rewrite this rule so it parses without warning? 
Or just ignore the warning? 

Thanks a lot! 

Have a great weekend! 

Michiel 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.antlr.org/pipermail/antlr-interest/attachments/20060710/58c77458/attachment-0001.html