[antlr-interest] [LEXER] Unwanted behaviour ?

[Gavin Lambert]

At 01:28 13/10/2009, Martin Potier wrote:
 >LO	: '[[';		// Link opening
 >LE	: ']]';		// Link ending
 >
 >PURETEXT
 >	:  ( ESC_SEQ |
 >~('\\'|TM1|TM2|TM3|TM4|ULM|LO|LE|OLM|BM|IM|UM|'|'|'\n') )+
 >	;

~ is a set operation; at the lexer level, you can only use single 
characters (or rules that result in simple alternatives of single 
characters).

Thus if you want to exclude '[['/']]' from the rule while keeping 
'['/']', you'll need to use this instead (assuming all the others 
are single characters):

PURETEXT
   : ( ESC_SEQ
     | '[' ~'['
     | ']' ~']'
     | ~('\\'|TM1|TM2|TM3|TM4|ULM|'['|']'|OLM|BM|IM|UM|'|'|'\n')
     )+
   ;