Antlr token priority

Question

I have a rule definition like this:

reference: volume':'first_page'-'last_page ;

volume: INTEGER;
first_page: INTEGER;
last_page: INTEGER;

INTEGER: [0-9]+;

FREE_TEXT_WORD: NON_SPACE+;

fragment NON_SPACE : ~[ \r\n\t];

Given the input "168:321-331", I thought it would match the reference rule. But in reality, the whole string is tokenized as a FREE_TEXT_WORD.

How can I make the INTEGER token take preference over FREE_TEXT_WORD in this case?

Thanks.

Answer 1

ANTLR will always use a longer token over a shorter token, so to correct this situation you must do one of the following things:

1. Make the FREE_TEXT_WORD not match more than 3 characters for the input 168:321-331, e.g. by not allowing it to contain a digit, or possibly removing the rule altogether.

   • You could also change FREE_TEXT_WORD to FREE_TEXT_CHARACTER. By limiting the rule to only matching a single character, it will never be longer than another token so its priority will be determined by its position in the grammar. You would then need to create a parser rule for words:

     freeTextWord : FREE_TEXT_CHARACTER+;
     

2. Move the FREE_TEXT_WORD token into a mode which is not enabled at the point where your input reaches 168:321-331.