I never worked with ANTLR and generative grammars, so this is my first attempt.
I have a custom language I need to parse.
Here's an example:
-- This is a comment
CMD.CMD1:foo_bar_123
CMD.CMD2
CMD.CMD4:9 of 28 (full)
CMD.NOTES:
This is an note.
A line
(1) there could be anything here foo_bar_123 & $ £ _ , . ==> BOOM
(3) same here
CMD.END_NOTES:
Briefly, there could be 4 types of lines:
1) -- comment
2) <section>.<command>
3) <section>.<command>: <arg>
4) <section>.<command>:
<arg1>
<arg2>
...
<section>.<end_command>:
<section> is the literal "CMD"
<command> is a single word (uppercase, lowercase letters, numbers, '_')
<end_command> is the same word of <command> but preceded by the literal "end_"
<arg> could be any character
Here's what I've done so far:
grammar MyGrammar;
/*
* Parser Rules
*/
root : line+ EOF ;
line : (comment_line | command_line | normal_line) NEWLINE;
comment_line : COMMENT ;
command_line : section '.' command ((COLON WHITESPACE*)? arg)? ;
normal_line : TEXT ;
section : CMD ;
command : WORD ;
arg : TEXT ;
/*
* Lexer Rules
*/
fragment LOWERCASE : [a-z] ;
fragment UPPERCASE : [A-Z] ;
fragment DIGIT : [0-9] ;
NUMBER : DIGIT+ ([.,] DIGIT+)? ;
CMD : 'CMD';
COLON : ':' ;
COMMENT : '--' ~[\r\n]*;
WHITESPACE : (' ' | '\t') ;
NEWLINE : ('\r'? '\n' | '\r')+;
WORD : (LOWERCASE | UPPERCASE | NUMBER | '_')+ ;
TEXT : ~[\r\n]* ;
This is a test for my grammar:
$antlr4 MyGrammar.g4
warning(146): MyGrammar.g4:45:0: non-fragment lexer rule TEXT can match the empty string
$javac MyGrammar*.java
$grun MyGrammar root -tokens
CMD.NEW
[#0,0:6='CMD.NEW',<TEXT>,1:0]
[#1,7:7='\n',<NEWLINE>,1:7]
[#2,8:7='<EOF>',<EOF>,2:0]
The problem is that "CMD.NEW" gets swallowed by TEXT, because that rule is greedy.
Anyone can help me with this?
Thanks
There is a grammar ambiguity.
In the example you have provided CMD.NEW can match both command_line and normal_line.
Thus, given the expression:
line : (comment_line | command_line | normal_line) NEWLINE;
the parser can not definitely say what rule to accept (command_line or normal_line), so it matches it to normal_line which is actually a simple TEXT.
Consider rewriting your grammar in the way the parser can always say what rule to accept.
UPDATE:
Try this (I did not test that, but it should work):
grammar MyGrammar;
/*
* Parser Rules
*/
root : line+ EOF ;
line : (comment_line | command_line) NEWLINE;
comment_line : COMMENT ;
command_line : CMD '.' (note_cmd | command);
command : command_name ((COLON WHITESPACE*)? arg)? ;
note_cmd : notes .*? (CMD '.' END_NOTES) ;
command_name : WORD ;
arg : TEXT ;
/*
* Lexer Rules
*/
fragment LOWERCASE : [a-z] ;
fragment UPPERCASE : [A-Z] ;
fragment DIGIT : [0-9] ;
NUMBER : DIGIT+ ([.,] DIGIT+)? ;
CMD : 'CMD';
COLON : ':' ;
COMMENT : '--' ~[\r\n]*;
WHITESPACE : (' ' | '\t') ;
NEWLINE : ('\r'? '\n' | '\r')+;
WORD : (LOWERCASE | UPPERCASE | NUMBER | '_')+ ;
NOTES : 'NOTES';
END_NOTES : 'END_NOTES';
TEXT : ~[\r\n]* ;
Related
This is freaking me out, I just can't find a solution to it. I have a grammar for search queries and would like to match any searchterm in a query composed out of printable letters except for special characters "(", ")". Strings enclosed in quotes are handled separately and work.
Here is a somewhat working grammar:
/* ANTLR Grammar for Minidb Query Language */
grammar Mdb;
start
: searchclause EOF
;
searchclause
: table expr
;
expr
: fieldsearch
| searchop fieldsearch
| unop expr
| expr relop expr
| lparen expr relop expr rparen
;
lparen
: '('
;
rparen
: ')'
;
unop
: NOT
;
relop
: AND
| OR
;
searchop
: NO
| EVERY
;
fieldsearch
: field EQ searchterm
;
field
: ID
;
table
: ID
;
searchterm
:
| STRING
| ID+
| DIGIT+
| DIGIT+ ID+
;
STRING
: '"' ~('\n'|'"')* ('"' )
;
AND
: 'and'
;
OR
: 'or'
;
NOT
: 'not'
;
NO
: 'no'
;
EVERY
: 'every'
;
EQ
: '='
;
fragment VALID_ID_START
: ('a' .. 'z') | ('A' .. 'Z') | '_'
;
fragment VALID_ID_CHAR
: VALID_ID_START | ('0' .. '9')
;
ID
: VALID_ID_START VALID_ID_CHAR*
;
DIGIT
: ('0' .. '9')
;
/*
NOT_SPECIAL
: ~(' ' | '\t' | '\n' | '\r' | '\'' | '"' | ';' | '.' | '=' | '(' | ')' )
; */
WS
: [ \r\n\t] + -> skip
;
The problem is that searchterm is too restricted. It should match any character that is in the commented out NOT_SPECIAL, i.e., valid queries would be:
Person Name=%
Person Address=^%Street%%%$^&*#^
But whenever I try to put NOT_SPECIAL in any way into the definition of searchterm it doesn't work. I have tried putting it literally into the rule, too (commenting out NOT_SPECIAL) and many others things, but it just doesn't work. In most of my attempts the grammar just complained about extraneous input after "=" and said it was expecting EOF. But I also cannot put EOF into NOT_SPECIAL.
Is there any way I can simply parse every text after "=" in rule fieldsearch until there is a whitespace or ")", "("?
N.B. The STRING rule works fine, but the user ought not be required to use quotes every time, because this is a command line tool and they'd need to be escaped.
Target language is Go.
You could solve that by introducing a lexical mode that you'll enter whenever you match an EQ token. Once in that lexical mode, you either match a (, ) or a whitespace (in which case you pop out of the lexical mode), or you keep matching your NOT_SPECIAL chars.
By using lexical modes, you must define your lexer- and parser rules in their own files. Be sure to use lexer grammar ... and parser grammar ... instead of the grammar ... you use in a combined .g4 file.
A quick demo:
lexer grammar MdbLexer;
STRING
: '"' ~[\r\n"]* '"'
;
OPAR
: '('
;
CPAR
: ')'
;
AND
: 'and'
;
OR
: 'or'
;
NOT
: 'not'
;
NO
: 'no'
;
EVERY
: 'every'
;
EQ
: '=' -> pushMode(NOT_SPECIAL_MODE)
;
ID
: VALID_ID_START VALID_ID_CHAR*
;
DIGIT
: [0-9]
;
WS
: [ \r\n\t]+ -> skip
;
fragment VALID_ID_START
: [a-zA-Z_]
;
fragment VALID_ID_CHAR
: [a-zA-Z_0-9]
;
mode NOT_SPECIAL_MODE;
OPAR2
: '(' -> type(OPAR), popMode
;
CPAR2
: ')' -> type(CPAR), popMode
;
WS2
: [ \t\r\n] -> skip, popMode
;
NOT_SPECIAL
: ~[ \t\r\n()]+
;
Your parser grammar would start like this:
parser grammar MdbParser;
options {
tokenVocab=MdbLexer;
}
start
: searchclause EOF
;
// your other parser rules
My Go is a bit rusty, but a small Java test:
String source = "Person Address=^%Street%%%$^&*#^()";
MdbLexer lexer = new MdbLexer(CharStreams.fromString(source));
CommonTokenStream tokens = new CommonTokenStream(lexer);
tokens.fill();
for (Token t : tokens.getTokens()) {
System.out.printf("%-15s %s\n", MdbLexer.VOCABULARY.getSymbolicName(t.getType()), t.getText());
}
print the following:
ID Person
ID Address
EQ =
NOT_SPECIAL ^%Street%%%$^&*#^
OPAR (
CPAR )
EOF <EOF>
Im writing chrome DEPS file parser. How to match one of either following grammar rule defintion of rightexpr. My grammar is like
the following one:
grammar Depsgrammar;
prog: expr+ EOF;
expr: varline
;
varline:
ID EQ rightexpr
;
rightexpr :
basicvalue | bentukonejsonval| bentuktwojsonval
;
bentukonejsonval :
'[' string? (COMMA string )* COMMA? ']'
;
bentuktwojsonval :
'{' singledictexpr? (COMMA singledictexpr )* COMMA? '}'
;
singledictexpr :
string ':' basicvalue
;
basicvalue :
True
| False
| string
| NUM
| varfunc
;
varfunc :
Var '(' string ')'
;
string :
SIMPLESTRINGEXPRDOUBLEQUOTE
| SIMPLESTRINGEXPRSINGLEQUOTE
;
Var : 'Var' ;
COMMA : ',' ;
NUM : [0-9]+;
ID : [a-zA-Z0-9_]+;
True : [tT] [Rr] [Uu] [Ee];
False: [Ff] [Aa] [Ll] [Ss] [Ee];
fragment SIMPLESTRINGEXPRDOUBLEQUOTEBASE : ~ ( '\n' | '\r' | '"' )* ;
SIMPLESTRINGEXPRDOUBLEQUOTE: '"' SIMPLESTRINGEXPRDOUBLEQUOTEBASE '"' ;
fragment SIMPLESTRINGEXPRSINGLEQUOTEBASE : ~ ( '\n' | '\r' | '\'' )* ;
SIMPLESTRINGEXPRSINGLEQUOTE : '\'' SIMPLESTRINGEXPRSINGLEQUOTEBASE '\'' ;
EQ : '=';
COMMENT:
'#' ~ ( '\n' | '\r' )* '\n' -> skip ;
WS : [ \n\t\r]+ -> skip ;
I want user could enter this input
#adas21 #FS;SFD33
_as= Var('das') # somelongth comment
_as_0= FALSE # somelongth comment
_as_0= 'as!' # somelongth comment
gclient_gn_args = [
#ad as!~;
'checkout_libaom',
'checkout_nacl',
'"{cros_board}" == "amd64-generic"',
'checkout_oculus_sdk',
]
vars = {
'checkout_libaom':1,
'checkout_nacl': "SS",
'checkout_oculus_sdk': FalSe,
'checkout_oculus_sdk':'',
}
s=[
]
whenever I enter simple syntax in grun
sa=true
always give me line 1:3 mismatched input 'true' expecting blah..(rightexpr def). I'm missing in understanding of basic antlr4 choice matching decision. Could you please teach me?
Thanks
Whenever you have an error where the list of expected tokens seemingly includes the unexpected token, it is a good idea to list the generated tokens. You can do that by passing the -tokens option to grun. If you do this for your input, you'll see that true is interpreted as an ID token, not a True token.
The reason for that is that when multiple lexer rules would match on the current input and produce a match of the same size, the one that's defined earlier in the grammar is chosen. So because ID is defined before True, it takes precedence. Generally all keywords should be defined before the ID rule to prevent exactly this issue.
In other words, moving the True and False rules before ID will solve your issue.
I am trying to parse ISO 8601 period expressions like "P3M2D", using antlr4. But I am hitting some kind of roadblock and will appreciate help. I am rather new to both antlr and compilers.
My grammar is as below. I have combined the lexer and parser rules in one go here:
grammar test_iso ;
// import testLexerRules ;
iso : ( date_expr NEWLINE)* EOF;
date_expr
: date_expr op=( '+' | '-' ) iso8601_interval #dateexpr_Interval
| date_expr op='-' date_expr #dateexpr_Diff
| DATETIME_NAME #dateexpr_Named
| '(' inner=date_expr ')' #dateexpr_Paren
;
///////////////////////////////////////////
iso8601_interval
: iso8601_interval_d
{ System.out.println("ISO8601_INTERVAL DATE seen " + $text);}
;
iso8601_interval_d
: 'P' ( y=NUMBER_INT 'Y' )? ( m=NUMBER_INT 'M' )? ( w=NUMBER_INT 'W' )? ( d=NUMBER_INT 'D' )?
;
///////////////////////////////////////////
// in separate file : test_lexer.g4
// lexer grammar testLexerRules ;
///////////////////////////////////////////
fragment
TODAY
: 'today' | 'TODAY'
;
fragment
NOW
: 'now' | 'NOW'
;
DATETIME_NAME
: TODAY
| NOW
;
///////////////////////////////////////////
NUMBER_INT
: '-'? INT // -3, 45
;
fragment
DIGIT : [0-9] ;
fragment
INT : '0' | [1-9] DIGIT* ;
//////////////////////////////////////////////
//
// identifiers
//
ID
: ALPHA ALPH_NUM*
{ System.out.println("ID seen " + getText()); }
;
ID_SQLFUNC
: 'h$' ALPHA_UPPER ALPHA_UPPER_NUM*
{ System.out.println("SQL FUNC seen " + getText()); }
;
fragment
ALPHA : [a-zA-Z] ;
fragment
ALPH_NUM : [a-zA-Z_0-9] ;
fragment
ALPHA_UPPER : [A-Z] ;
fragment
ALPHA_UPPER_NUM : [A-Z_0-9] ;
//////////////////////////////////////////////
NEWLINE : '\r\n' ;
WS : [ \t]+ -> skip ;
In test run, it never hits the iso8601_interval_d rule, it always goes to ID rule.
C:\lab>java org.antlr.v4.gui.TestRig test_iso iso -tokens -tree
now + P3M2D
^Z
ID seen P3M2D
[#0,0:2='now',<DATETIME_NAME>,1:0]
[#1,4:4='+',<'+'>,1:4]
[#2,6:10='P3M2D',<ID>,1:6]
[#3,11:12='\r\n',<'
'>,1:11]
[#4,13:12='<EOF>',<EOF>,2:0]
line 1:6 mismatched input 'P3M2D' expecting 'P'
ISO8601_INTERVAL DATE seen P3M2D
(iso (date_expr (date_expr now) + (iso8601_interval (iso8601_interval_d P3M2D))) \r\n <EOF>)
If I remove the "ID" rule and run again, it parses as desired:
now + P3M2D
^Z
[#0,0:2='now',<DATETIME_NAME>,1:0]
[#1,4:4='+',<'+'>,1:4]
[#2,6:6='P',<'P'>,1:6]
[#3,7:7='3',<NUMBER_INT>,1:7]
[#4,8:8='M',<'M'>,1:8]
[#5,9:9='2',<NUMBER_INT>,1:9]
[#6,10:10='D',<'D'>,1:10]
[#7,11:12='\r\n',<'
'>,1:11]
[#8,13:12='<EOF>',<EOF>,2:0]
ISO8601_INTERVAL DATE seen P3M2D
(iso (date_expr (date_expr now) + (iso8601_interval (iso8601_interval_d P 3 M 2 D))) \r\n <EOF>)
I also tried prefixing a special character like "#" in the parser rule
iso8601_interval_d
: '#P' ( y=NUMBER_INT 'Y' )? ( m=NUMBER_INT 'M' )? ( w=NUMBER_INT 'W' )? ( d=NUMBER_INT 'D' )?
;
but now a different kind of failure
now + #P3M2D
^Z
ID seen M2D
[#0,0:2='now',<DATETIME_NAME>,1:0]
[#1,4:4='+',<'+'>,1:4]
[#2,6:7='#P',<'#P'>,1:6]
[#3,8:8='3',<NUMBER_INT>,1:8]
[#4,9:11='M2D',<ID>,1:9]
[#5,12:13='\r\n',<'
'>,1:12]
[#6,14:13='<EOF>',<EOF>,2:0]
line 1:9 no viable alternative at input '3M2D'
ISO8601_INTERVAL DATE seen #P3M2D
(iso (date_expr (date_expr now) + (iso8601_interval (iso8601_interval_d #P 3 M2D))) \r\n <EOF>)
I am sure I am not the first one to hit upon something like this. What is the antlr idiom here?
EDIT -- I need the ID token elsewhere in other parts of my grammar that I have omitted here, to highlight the problem I am facing.
Like find out even by other, the issue is in the ID token. The fact is that the duration syntax for iso-8601 is a valid ID. Besides the solution figured out by #Mike. If something called island grammar is suitable for your needs you can use ANTLR's lexical modes to exclude the ID lexer rule while parsing an iso date.
Belove there is an examples on how it could work
parser grammar iso;
options { tokenVocab=iso_lexer; }
iso : ISO_BEGIN ( date_expr NEWLINE)* ISO_END;
date_expr
: date_expr op=( '+' | '-' ) iso8601_interval #dateexpr_Interval
| date_expr op='-' date_expr #dateexpr_Diff
| DATETIME_NAME #dateexpr_Named
| '(' inner=date_expr ')' #dateexpr_Paren
;
///////////////////////////////////////////
iso8601_interval
: iso8601_interval_d
{ System.out.println("ISO8601_INTERVAL DATE seen " + $text);}
;
iso8601_interval_d
: 'P' ( y=NUMBER_INT 'Y' )? ( m=NUMBER_INT 'M' )? ( w=NUMBER_INT 'W' )? ( d=NUMBER_INT 'D' )?
;
then in the lexer
lexer grammar iso_lexer;
//
// identifiers (in DEFAULT_MODE)
//
ISO_BEGIN
: '<#' -> mode(ISO)
;
ID
: ALPHA ALPH_NUM*
{ System.out.println("ID seen " + getText()); }
;
ID_SQLFUNC
: 'h$' ALPHA_UPPER ALPHA_UPPER_NUM*
{ System.out.println("SQL FUNC seen " + getText()); }
;
WS0 : [ \t]+ -> skip ;
// all the following token are scanned only when iso mode is active
mode ISO;
ISO_END
: '#>' -> mode(DEFAULT_MODE)
;
WS0 : [ \t]+ -> skip ;
NEWLINE : '\r'? '\n' ;
ADD : '+' ;
SUB : '-' ;
LPAREN : '(' ;
RPAREN : ')' ;
P : 'P' ;
Y : 'Y' ;
M : 'M' ;
W : 'W' ;
D : 'D' ;
DATETIME_NAME
: TODAY
| NOW
;
fragment TODAY: 'today' | 'TODAY' ;
fragment NOW : 'now' | 'NOW' ;
///////////////////////////////////////////
NUMBER_INT
: '-'? INT // -3, 45
;
fragment DIGIT : [0-9] ;
fragment INT : '0' | [1-9] DIGIT* ;
//////////////////////////////////////////////
fragment ALPHA : [a-zA-Z] ;
fragment ALPH_NUM : [a-zA-Z_0-9] ;
fragment ALPHA_UPPER : [A-Z] ;
fragment ALPHA_UPPER_NUM : [A-Z_0-9] ;
Such grammar can parse expressions like
Pluton Planet <% now + P10Y
%>
I changed a bit the parser rule iso to demonstrate ID and period mixing.
Hope this helps
It's not possible what you wanna do. ID matches the same input as iso8601_interval. In such cases ANTLR4 picks the longest match, which is ID as it can match an unlimited number of characters.
The only way you could possible make it work in the grammar is to exclude P as a possible ID introducer, which then can exclusively be used for the duration.
Another option is a post processing step. Parse durations like any other identifier and in your semantic phase check all those ids that look like a duration. This is probably the best solution.
Background
I have been using ANTLRWorks (V 1.4.3) for a few days now and trying to write a simple Boolean parser. The combined lexer/parser grammar below works well for most of the requirements including support for quoted white-spaced text as operands for a Boolean expression.
Problem
I would like the grammar to work for white-spaced operands without the need of quotes.
Example
For example, expression-
"left right" AND center
should have the same parse tree even after dropping the quotes-
left right AND center.
I have been learning about backtracking, predicates etc but can't seem to find a solution.
Code
Below is the grammar I have got so far. Any feedback on the foolish mistakes is appreciated :).
Lexer/Parser Grammar
grammar boolean_expr;
options {
TokenLabelType=CommonToken;
output=AST;
ASTLabelType=CommonTree;
}
#modifier{public}
#ctorModifier{public}
#lexer::namespace{Org.CSharp.Parsers}
#parser::namespace{Org.CSharp.Parsers}
public
evaluator
: expr EOF
;
public
expr
: orexpr
;
public
orexpr
: andexpr (OR^ andexpr)*
;
public
andexpr
: notexpr (AND^ notexpr)*
;
public
notexpr
: (NOT^)? atom
;
public
atom
: word | LPAREN! expr RPAREN!
;
public
word
: QUOTED_TEXT | TEXT
;
/*
* Lexer Rules
*/
LPAREN
: '('
;
RPAREN
: ')'
;
AND
: 'AND'
;
OR
: 'OR'
;
NOT
: 'NOT'
;
WS
: ( ' ' | '\t' | '\r' | '\n') {$channel=HIDDEN;}
;
QUOTED_TEXT
: '"' (LETTER | DIGIT | ' ' | ',' | '-')+ '"'
;
TEXT
: (LETTER | DIGIT)+
;
/*
Fragment lexer rules can be used by other lexer rules, but do not return tokens by themselves
*/
fragment DIGIT
: ('0'..'9')
;
fragment LOWER
: ('a'..'z')
;
fragment UPPER
: ('A'..'Z')
;
fragment LETTER
: LOWER | UPPER
;
Simply let TEXT in your atom rule match once or more: TEXT+. When it matches a TEXT token more than once, you'll also want to create a custom root node for these TEXT tokens (I added an imaginary token called WORD in the grammar below).
grammar boolean_expr;
options {
output=AST;
}
tokens {
WORD;
}
evaluator
: expr EOF
;
...
word
: QUOTED_TEXT
| TEXT+ -> ^(WORD TEXT+)
;
...
Your input "left right AND center" would now be parsed as follows:
Below is a cut down version of a grammar that is parsing an input assembly file. Everything in my grammar is fine until i use labels that have 3 characters (i.e. same length as an OPCODE in my grammar), so I'm assuming Antlr is matching it as an OPCODE rather than a LABEL, but how do I say "in this position, it should be a LABEL, not an OPCODE"?
Trial input:
set a, label1
set b, abc
Output from a standard rig gives:
line 2:5 missing EOF at ','
(OP_BAS set a (REF label1)) (OP_SPE set b)
When I step debug through ANTLRWorks, I see it start down instruction rule 2, but at the reference to "abc" jumps to rule 3 and then fail at the ",".
I can solve this with massive left factoring, but it makes the grammar incredibly unreadable. I'm trying to find a compromise (there isn't so much input that the global backtrack is a hit on performance) between readability and functionality.
grammar TestLabel;
options {
language = Java;
output = AST;
ASTLabelType = CommonTree;
backtrack = true;
}
tokens {
NEGATION;
OP_BAS;
OP_SPE;
OP_CMD;
REF;
DEF;
}
program
: instruction* EOF!
;
instruction
: LABELDEF -> ^(DEF LABELDEF)
| OPCODE dst_op ',' src_op -> ^(OP_BAS OPCODE dst_op src_op)
| OPCODE src_op -> ^(OP_SPE OPCODE src_op)
| OPCODE -> ^(OP_CMD OPCODE)
;
operand
: REG
| LABEL -> ^(REF LABEL)
| expr
;
dst_op
: PUSH
| operand
;
src_op
: POP
| operand
;
term
: '('! expr ')'!
| literal
;
unary
: ('+'! | negation^ )* term
;
negation
: '-' -> NEGATION
;
mult
: unary ( ( '*'^ | '/'^ ) unary )*
;
expr
: mult ( ( '+'^ | '-'^ ) mult )*
;
literal
: number
| CHAR
;
number
: HEX
| BIN
| DECIMAL
;
REG: ('A'..'C'|'I'..'J'|'X'..'Z'|'a'..'c'|'i'..'j'|'x'..'z') ;
OPCODE: LETTER LETTER LETTER;
HEX: '0x' ( 'a'..'f' | 'A'..'F' | DIGIT )+ ;
BIN: '0b' ('0'|'1')+;
DECIMAL: DIGIT+ ;
LABEL: ( '.' | LETTER | DIGIT | '_' )+ ;
LABELDEF: ':' ( '.' | LETTER | DIGIT | '_' )+ {setText(getText().substring(1));} ;
STRING: '\"' .* '\"' {setText(getText().substring(1, getText().length()-1));} ;
CHAR: '\'' . '\'' {setText(getText().substring(1, 2));} ;
WS: (' ' | '\n' | '\r' | '\t' | '\f')+ { $channel = HIDDEN; } ;
fragment LETTER: ('a'..'z'|'A'..'Z') ;
fragment DIGIT: '0'..'9' ;
fragment PUSH: ('P'|'p')('U'|'u')('S'|'s')('H'|'h');
fragment POP: ('P'|'p')('O'|'o')('P'|'p');
The parser has no influence on what tokens the lexer produces. So, the input "abc" will always be tokenized as a OPCODE, no matter what the parser tries to match.
What you can do is create a label parser rules that matches either a LABEL or OPCODE and then use this label rule in your operand rule:
label
: LABEL
| OPCODE
;
operand
: REG
| label -> ^(REF label)
| expr
;
resulting in the following AST for your example input:
This will only match OPCODE, but will not change the type of the token. If you want the type to be changed as well, add a bit of custom code to the rule that changes it to type LABEL:
label
: LABEL
| t=OPCODE {$t.setType(LABEL);}
;