Rule matching either some literals or specific words followed by numbers

[just-ero]

Hello,

sorry, I didn't know how else to explain it.

What I would like to make work is the following:

fragment LENGTH: [0-9]+;

STRING: 'string' LENGTH;
BYTE_ARRAY: 'byte' LENGTH;

type :
    'bool'
    | 'sbyte'
    | 'byte'
    | 'short'
    | 'ushort'
    | 'int'
    | 'uint'
    | 'long'
    | 'ulong'
    | 'float'
    | 'double'
    | STRING
    | BYTE_ARRAY
    ;

However, currently, I get an error saying string32 is unexpected and it lists every member except for string[0-9]+ and byte[0-9]+.
How can I make this work?

Additionally, I seem to have a whitespace problem: mismatched input '{\r\n}' expecting '{'
For this rule: state : 'state' '(' STRING (',' STRING)? ')' '{' declaration+ '}'

Why would this happen?

[kaby76]

There isn't enough information to form an answer. We need the entire grammar and input.

[just-ero]

I had some things declared as parser rules when they should have been lexer rules. This was my whole problem.

Is there a document somewhere outline when something should be a a lexer rule or a parser rule? And when we should use fragments?

[kaby76]

• There are no hard rules for where to draw the boundary between lexer and parser. You could even define everything except a single character in the parser, but would likely suffer in performance.
• Study how it's done with other grammars. See grammars-v4.
• Use fragment when you want to create a reusable regular expression but where you don't want to generate a token.
• Make sure to understand the basic rules of how Antlr lexers work. See https://stackoverflow.com/questions/78856126/odata-parser-problem-with-substring-function#comment139034947_78856126
• Read StackOverflow.com answers for Antlr questions.