为什么这样解析

Question

I have the following grammar:

grammar Demo;

program: command
         IDENTIFIER
         ;

command:
       | 'add'
       | 'remove'
       ;

IDENTIFIER: [a-zA-Z][a-zA-Z0-9]* ;

WHITESPACE: [ \t\n\r]+ -> skip;

Now I can entered something like "add foo" and I get the right result. But he also accept "foo" only. I thought that the parser will throw an exception, because the value have to start with one of the commands? Is there an option to fix the problem? Or have I a fallacy?

Answer 1

As indicated in the comments by Seelenvirtuose, the problem is that your rule for command allows an empty command, it should be:

command:
         'add'
       | 'remove'
       ;

The pipe symbol ( | ) is a separator between alternatives, in your original code it separated an empty production from 'add' .

However, as is, the grammar would still allow to only match foo which is because you don't have an explicit EOF token in your program rule (similar to the $ option in a regular expression). Without the EOF token the parser happily matches what it can and ignores the rest. So, if that is no the desired behavior always end your main rule with EOF:

program: command IDENTIFIER EOF;