[antlr-interest] StringTemplate question

[Gabriel Miro]

Hello all,

I'm writing a translator to add tab characters before specific elements of a
string, e.g.:

Input : (FIELD1 + FIELD2) > 'VALUE' OR FIELD3 like ('%somevalue%')
output:
\t(\tFIELD1\t+\tFIELD2)\t>\t'VALUE'\tOR\tFIELD3\tlike\t(\t'%somevalue%'\t)

and I'm thinking about the best way to do that. I already wrote the parser
and I'm working on the translator/emitter.
For instance, in the grammar below I find that the template from 'expr' gets
overriden with the template of the last matched sub rule. Is there a way to
merge the subrule templates into the main rule one? Also, does the template
declaration need to be in the first rule?

grammar Test;

options{
    output=template;
}

rule    :    e+=expr+ -> template(exps={$e}) "<exps>";
expr    :    (B_OP WS?)* lside WS* C_OP WS* STRING
            -> template(bop={$B_OP.text},ls={lside.text},
                op={$C_OP.text},val={$STRING.text}) "    <bop>    <ls>
<op>    <val>";
lside    :
        FIELD
            -> template(f={$FIELD.text}) "    <f>"
        |
        LPAREN WS* alside WS* A_OP WS* arside WS* RPAREN
            -> template(ls={$alside.text},op={$A_OP.text},rs={$arside.text})
"    (    <ls>    <op>    <rs>    )";
alside    :    FIELD;
arside    :    FIELD;
FIELD    :    'A'..'Z'+;
RPAREN    :    ')';
LPAREN    :    '(';
A_OP    :    '+'|'-';
C_OP    :    '<>'|'='|'<'|'>'|'<='|'>=';
B_OP    :    'AND'|'OR';
LIKE    :    'like' LPAREN STRING* RPAREN;

WS  :   ( ' '
        | '\t'
        ) {$channel=HIDDEN;}
    ;

STRING
    :  '\'' ( ESC_SEQ | ~('\\'|'\'') )* '\''
    ;

fragment
HEX_DIGIT : ('0'..'9'|'a'..'f'|'A'..'F') ;

fragment
ESC_SEQ
    :   '\\' ('b'|'t'|'n'|'f'|'r'|'\"'|'\''|'\\')
    |   UNICODE_ESC
    |   OCTAL_ESC
    ;

fragment
OCTAL_ESC
    :   '\\' ('0'..'3') ('0'..'7') ('0'..'7')
    |   '\\' ('0'..'7') ('0'..'7')
    |   '\\' ('0'..'7')
    ;

fragment
UNICODE_ESC
    :   '\\' 'u' HEX_DIGIT HEX_DIGIT HEX_DIGIT HEX_DIGIT
    ;