[Question] Why is the ATN of the following lexer rules different from the javadoc of ATNState

I have the following toy lexer

fragment DIGIT
    : [0-9]
    ;

NaturalNumber
    : DIGIT+
    ;

According to the javadoc of ATNState, the ATN of NaturalNumber should start with RuleStartState --> PlusBlockStartState.

But In fact, a path in the ATN looks similar to

RuleStartState -> BasicState -> SetTransition[0-9] ->RuleStopState -> BlockEndState 
          ^                                                               |
          |                                                               |
  BasicState <----------------------  PlusBlockStartState <- ... <- PlusLoopbackState

Could someone advise why the ATN is changed in this way, and is it possible to get a canonical ATN as described in the javadoc of ATNState?

Thanks in advance.

[sharwell]

It seems like it would be better to convert this to a GitHub discussion instead of moving it to a completely different medium.

[chengniansun]

tate tran

So, for the lexer rule NaturalNumber, there will be no RuleStopState which has no outgoing transitions.