Question

Consider the following simplified grammar:

ID  :   ('a'..'z') ('a'..'z')*;

TYPE    :   'int' '[' ']' |
        'boolean' |
        'int';

Now if I write boolean as input, it can be parsed as ID or as TYPE. I want it to be parsed as TYPE – how?

Answer 1

Put TYPE before ID:

TYPE : 'int' '[' ']' 
     | 'boolean' 
     | 'int'
     ;

ID   : ('a'..'z') ('a'..'z')*
     ;

ANTLR’s lexer rules work like this:

1. match as much as possible
2. if after rule #1, two or more rules match the same amount of characters, pick the rule defined first in the grammar

Because of these rules, "boolean" is tokenized as TYPE, while "booleans" would be tokenized as ID ¹

---

¹ "booleans" is not tokenized as two tokens: TYPE ("boolean") and an ID ("s"))