Trying to create a grammar for schemas

[rowan-vr]

Hello! I'm trying to make a grammar to parse easy schema in a structure like the prisma orm uses. After trying a lot of different things it still doesn't parse it properly and I am not sure where the issue lies.

This is the lexer I have written:

lexer grammar OrmLexer;

MODEL: 'model';
MODELNAME : LETTERS+;
OPEN : '{' NEWLINE?;

OPENMODEL   : MODEL MODELNAME OPEN -> pushMode(MODELMODE) ;

WHITESPACE : ' ' -> skip ;
NEWLINE: '\r' '\n' | '\n' | '\r';

mode MODELMODE ;
FIELD: FIELDNAME FIELDTYPE NEWLINE;
FIELDNAME   : LETTERS+;
FIELDTYPE   : LETTERS+;
CLOSE: '}' -> popMode ;

fragment LETTERS  : [a-zA-Z] ;

This is the parser I have written:

parser grammar OrmParser;

options { tokenVocab=OrmLexer; }

start: root | EOF ;
root: model*;

model : modelstart modelbody modelend;
modelbody: modelfield*;
modelstart : MODEL MODELNAME OPEN ;
modelfield : FIELDNAME FIELDTYPE NEWLINE ;
modelend: CLOSE ;

I am testing the grammar on an example schema which looks like:

model test {
    id test
    user test
}

In this schema the fields are not being parsed correctly, there should be two fields namely:
(FIELDNAME: id, FIELDTYPE: test) and (FIELDNAME: user, FIELDTYPE: test) however it doesn't parse those, this is the tree I get in result:

Besides that I am not really sure when to use fragments and when not in the lexer.

I hope anyone can help me in finding/solving the issue! Thanks in advance