ANTLR struggling to parse integer vs quoted string

Question

I'm trying to create a language using ANTLR where each line consists of an instruction, where an instruction is an opcode and any number of operands like so:

aaa "str1" "str2" 123
bbb 123 "str" 456
ccc
ddd

I have strings seemingly working OK, but integers seem to be parsed incorrectly.

Here's my complete grammar file:

grammar Insn;

prog: (line? NEWLINE)+;

line: instruction;
instruction: instruction_name instruction_operands?;

instruction_name: IDENTIFIER;
instruction_operands: instruction_operand instruction_operand*;
instruction_operand: ' '+ (operand_int | operand_string);

operand_int: INT;
operand_string: QSTRING;

NEWLINE : [\r\n]+;
IDENTIFIER: [a-zA-Z0-9_\-]+;
INT: '-'?[0-9]+;
QSTRING: '"' (~('"' | '\\' | '\r' | '\n') | '\\' ('"' | '\\'))* '"';
COMMENT: ';' ~[\r\n]* -> channel(HIDDEN);

I've tried multiple different INT definitions such as INT: '-'?('0'..'9')+; and INT: '2'; making all the INTs in the input 2, always resulting in an error similar to line 1:18 extraneous input '123' expecting {' ', INT, QSTRING}, with the line number, column and 123 integer replaced with whatever it was parsing.

Here's the parse tree generated by ANTLR's tooling as used in the ANTLR getting-started.md document.

I'm completely new to ANTLR and am not familiar with lots of terminology so please keep it simple for me.

Answer 1

The problem is that 123is recognised as IDENTIFIER because it is a valid identifier (all INTs are). Both of them must be distinguishable. IDENTIFIER should probably be something like this IDENTIFIER: [a-zA-Z][a-zA-Z0-9_\-]*;