Geek asks and answers

ANTLR, Problem with grammar expressions

I recently started using ANTLR. I am currently trying to code grammar expressions using the +, -, *and array[index]and several other structures.

This is the desired grammar:

Exp -> Exp (+ | - | * | < | &&) Exp
     | Exp [ Exp ]
     | -Exp
     | ( Exp )
     | Exp.length
     | true
     | false
     | Id
     | this
     | ! Exp

At first, I reorganized it AndExp, SumExp, ProdExpand so on, to eliminate the priority. Like that:

Exp        -> AndExp
AndExp     -> CmpExp (&& CmpExp)*
CmpExp     -> SumExp (< SumExp)*
SumExp     -> ProdExp ((+|-) ProdExp)*
ProdExp    -> UnaryExp (Times UnaryExp)*
UnaryExp   -> Minus* PrimaryExp
PrimaryExp -> Exp.length
            | Exp [ Exp ]
            | ! Exp
            | true
            | false
            | Id
            | this

Then I realized that this uses left recursion, and that ANTLR doesn't like this. I went on to eliminate left recursion , and I ended up with this grammar:

grammar test;

options {
    language=Java;
    output=AST;
    backtrack=true;
}

start      : expression;

expression : andExp;
andExp     : cmpExp (And^ cmpExp)*;
cmpExp     : sumExp (LessThan^ sumExp)*;
sumExp     : prodExp ((Plus | Minus)^ prodExp)*;
prodExp    : unaryExp (Times^ unaryExp)*;
unaryExp   : Minus* primaryExp;

primaryExp : INT                   primaryPrime
           | 'true'                primaryPrime
           | 'false'               primaryPrime
           | 'this'                primaryPrime
           | ID                    primaryPrime
           | '!' expression        primaryPrime
           | '('! expression ')'!  primaryPrime
           ;


primaryPrime
           : '[' expression ']'             primaryPrime
           | '.' ID '(' exprList ')'        primaryPrime
           | '.length'                      primaryPrime
           | 'new' 'int' '[' expression ']' primaryPrime
           | 'new' ID '(' ')'               primaryPrime
           |
           ;


exprList   :
           | expression (',' expression)*;

LessThan   : '<';
Plus       : '+';
Minus      : '-';
Times      : '*';
And        : '&&';
Not        : '!';
INT        :    '0' | ('1'..'9')('0'..'9')*;
ID         :    ('a'..'z'|'A'..'Z'|'_') ('a'..'z'|'A'..'Z'|'0'..'9'|'_')*;
WS         : ('\t' | ' ' | '\r' | '\n'| '\u000C') { $channel=HIDDEN; } ;

  • Q1: Tracking is “required” for this type of grammar (I can’t get it through ANTLR if I don’t activate it), or did I miss something simple?

  • Q2: ^ -> ^(TOKEN ...), , (- primaryPrime, ):

    primaryPrime
    : '[' expression ']' primaryPrime -> ^(ARRAY_EXPR expression)
//...

    array[index]

    array
ARRAY_EXPR
    index


    ARRAY_EXPR
    array
    index

    ? , ?

+3
1

A1

, (). , backtrack=true , , . backtrack=true, backtracking , , , . , , backtrack=true , , , , . , .

, , . -, . :

foo : bar baz ;
bar : 'bar' ;
baz : 'baz' | /* epsilon */ ;


foo : bar baz? ;
bar : 'bar' ;
baz : 'baz' ;

.

, , true, false .., : lexer. Lexer , ( ) , ID, . .

A2

, , () .

:

grammar test;

options {
  output=AST;
}

tokens {
  ARRAY_EXPR;
}

start      : expression;

expression : andExp;
andExp     : cmpExp (And^ cmpExp)*;
cmpExp     : sumExp (LessThan^ sumExp)*;
sumExp     : prodExp ((Plus | Minus)^ prodExp)*;
prodExp    : unaryExp (Times^ unaryExp)*;
unaryExp   :  '-' primaryExp
           |  '!' primaryExp // negation is really a `unaryExp`
           |  primaryExp
           ;

primaryExp : INT                  primaryPrime[null]?
           | 'true'               primaryPrime[null]?
           | 'false'              primaryPrime[null]?
           | 'this'               primaryPrime[null]?
           | (ID -> ID)           (primaryPrime[new CommonTree($ID)] -> primaryPrime)?
           | '('! expression ')'! primaryPrime[null]?
           ;

// removed the matching of 'epsilon'
primaryPrime [CommonTree parent]
           : '[' expression ']'             primaryPrime[null]? -> ^(ARRAY_EXPR {parent} expression primaryPrime?)
           | '.' ID '(' exprList? ')'       primaryPrime[null]?
           | '.length'                      primaryPrime[null]?
           | 'new' 'int' '[' expression ']' primaryPrime[null]?
           | 'new' ID '(' ')'               primaryPrime[null]?
           ;

// removed the matching of 'epsilon' 
exprList   : expression (',' expression)*;

// be sure to create explicit tokens for keywords!
True       : 'true';
False      : 'false';
This       : 'this';
LessThan   : '<';
Plus       : '+';
Minus      : '-';
Times      : '*';
And        : '&&';
Not        : '!';
INT        : '0' | ('1'..'9')('0'..'9')*;
ID         : ('a'..'z'|'A'..'Z'|'_') ('a'..'z'|'A'..'Z'|'0'..'9'|'_')*;
WS         : ('\t' | ' ' | '\r' | '\n'| '\u000C') { $channel=HIDDEN; } ;

"array[2*3]" AST:

enter image description here

, :

import org.antlr.runtime.*;
import org.antlr.runtime.tree.*;
import org.antlr.stringtemplate.*;

public class Main {
  public static void main(String[] args) throws Exception {
    String source = "array[2*3]";
    testLexer lexer = new testLexer(new ANTLRStringStream(source));
    CommonTokenStream tokens = new CommonTokenStream(lexer);
    testParser parser = new testParser(tokens);
    testParser.start_return returnValue = parser.start();
    CommonTree tree = (CommonTree)returnValue.getTree();
    DOTTreeGenerator gen = new DOTTreeGenerator();
    StringTemplate st = gen.toDOT(tree);
    System.out.println(st);
  }
}
+4

More articles:


All Articles