I am trying to write a visitor for a simple antlr4 grammar - I am adapting from the following example from the book:
* directory tour
* example: LabeledExpr.g4, EvalVisitor.java, Calc.java
Based on the java code, I have written the following go code:
package main
import (
"os"
"./parser"
"github.com/antlr/antlr4/runtime/Go/antlr"
)
type evalVisitor struct {
*parser.BaseLabeledExprVisitor
}
func (v *evalVisitor) VisitAddSub(c *parser.AddSubContext) int {
left := v.Visit(c.Expr(0))
right := v.Visit(c.Expr(1))
if(c.GetOp().GetTokenType() == parser.LabeledExprParserADD) {
return left + right //error: invalid operation: left + right (operator + not defined on interface)
} else {
return left - right
}
}
func main() {
input, _ := antlr.NewFileStream(os.Args[1])
lexer := parser.NewLabeledExprLexer(input)
stream := antlr.NewCommonTokenStream(lexer, antlr.TokenDefaultChannel)
p := parser.NewLabeledExprParser(stream)
tree := p.Prog()
var visitor evalVisitor
visitor.Visit(tree)
}
I am showing one of the visitors above, the other visitors would be written similarly. I am getting some compile errors as shown in comments above. How to fix this error?
There also seems to be an error in the top level calling of the visitor, because when I commented out the "left+right" line, I got a SIGSEGV fault.
For your reference, I am showing below the original java code:
public Integer visitAddSub(LabeledExprParser.AddSubContext ctx) {
int left = visit(ctx.expr(0)); // get value of left subexpression
int right = visit(ctx.expr(1)); // get value of right subexpression
if ( ctx.op.getType() == LabeledExprParser.ADD ) return left + right;
return left - right; // must be SUB
}
Also, the grammar is like this:
grammar LabeledExpr;
prog: stat+ ;
stat: expr NEWLINE # printExpr
| ID '=' expr NEWLINE # assign
| NEWLINE # blank
;
expr: expr op=('*'|'/') expr # MulDiv
| expr op=('+'|'-') expr # AddSub
| INT # int
| ID # id
| '(' expr ')' # parens
;
MUL : '*' ; // assigns token name to '*' used above in grammar
DIV : '/' ;
ADD : '+' ;
SUB : '-' ;
ID : [a-zA-Z]+ ; // match identifiers
INT : [0-9]+ ; // match integers
NEWLINE:'\r'? '\n' ; // return newlines to parser (is end-statement signal)
WS : [ \t]+ -> skip ; // toss out whitespace
NOTE: I searched around for a sample visitor code, but I hit some negative comments at 54992660, and that is also posted on antlr issues. That question has an answer that is incomplete and does not compile. So, do visitors at all work in the Go target of antlr4? And is there a sample code available for that?
I Googled a bit, and hacked the following Go visitor together:
package antlr4demo
import (
"strconv"
"github.com/antlr/antlr4/runtime/Go/antlr"
)
type EvalVisitor struct {
BaseExpressionVisitor
Results map[int]float64
}
func (v *EvalVisitor) Visit(tree antlr.ParseTree) float64 {
switch val := tree.(type) {
case *ParseContext:
return v.VisitParse(val)
case *MultDivExprContext:
return v.VisitMultDivExpr(val)
case *NumberExprContext:
return v.VisitNumberExpr(val)
case *PlusSubExprContext:
return v.VisitPlusSubExpr(val)
case *NestedExprContext:
return v.VisitNestedExpr(val)
case *UnaryExprContext:
return v.VisitUnaryExpr(val)
default:
panic("Unknown context")
}
}
func (v *EvalVisitor) VisitParse(ctx *ParseContext) float64 {
for index, expr := range ctx.expr_list {
v.Results[index] = v.Visit(expr)
}
return v.Results[len(v.Results)-1]
}
func (v *EvalVisitor) VisitMultDivExpr(ctx *MultDivExprContext) float64 {
lhs := v.Visit(ctx.lhs)
rhs := v.Visit(ctx.rhs)
if ctx.op.GetTokenType() == ExpressionLexerMULT {
return lhs * rhs
} else {
return lhs / rhs
}
}
func (v *EvalVisitor) VisitPlusSubExpr(ctx *PlusSubExprContext) float64 {
lhs := v.Visit(ctx.lhs)
rhs := v.Visit(ctx.rhs)
if ctx.op.GetTokenType() == ExpressionLexerPLUS {
return lhs + rhs
} else {
return lhs - rhs
}
}
func (v *EvalVisitor) VisitNumberExpr(ctx *NumberExprContext) float64 {
val, _ := strconv.ParseFloat(ctx.NUMBER().GetText(), 10)
return val
}
func (v *EvalVisitor) VisitNestedExpr(ctx *NestedExprContext) float64 {
return v.Visit(ctx.Expr())
}
func (v *EvalVisitor) VisitUnaryExpr(ctx *UnaryExprContext) float64 {
return -v.Visit(ctx.Expr())
}
grammar Expression;
parse
: expr_list+=expr+ EOF
;
expr
: '(' expr ')' #NestedExpr
| SUB expr #UnaryExpr
| lhs=expr op=( MULT | DIV ) rhs=expr #MultDivExpr
| lhs=expr op=( PLUS | SUB ) rhs=expr #PlusSubExpr
| NUMBER #NumberExpr
;
MULT : '*';
DIV : '/';
PLUS : '+';
SUB : '-';
NUMBER
: ( D* '.' )? D+
;
SPACES
: [ \t\r\n] -> skip
;
fragment D : [0-9];
First download the ANTLR 4.9 JAR, generate the parser and visitor Go files and move them to the antlr4demo
folder:
wget https://www.antlr.org/download/antlr-4.9-complete.jar
java -cp antlr-4.9-complete.jar org.antlr.v4.Tool -Dlanguage=Go -o antlr4demo -package antlr4demo -visitor -no-listener Expression.g4
If you now run the following Go script:
package main
import (
"fmt"
"./antlr4demo"
"github.com/antlr/antlr4/runtime/Go/antlr"
)
func main() {
expression := "1000 25/5 (1 + 2) * -3.14159265"
input := antlr.NewInputStream(expression)
lexer := antlr4demo.NewExpressionLexer(input)
stream := antlr.NewCommonTokenStream(lexer, 0)
parser := antlr4demo.NewExpressionParser(stream)
parser.BuildParseTrees = true
tree := parser.Parse()
visitor := antlr4demo.EvalVisitor{
Results: make(map[int]float64),
}
var result = visitor.Visit(tree)
fmt.Println(expression, "=", result)
fmt.Println("All results: ", visitor.Results)
}
you'll see the output:
$ go run main.go
1000 25/5 (1 + 2) * -3.14159265 = -9.424777950000001
All results: map[0:1000 1:5 2:-9.424777950000001]
Note that I have never programed anything in Go: I'm sure the code is a mess, but hey, "it works".
You're welcome @R71, and thanks for the edit: forgot about the
-o
option.if the top level grammar rule was "parse:expr+", how would the code in Visit() and VisitParse() change? Basically, trying to understand how lists are represented in the tree and how to traverse them.
@R71 you could collect the expressions using ANTLR's
+=
operator and then loop through the list inside theVisitParse
method of the visitor. I updated my answer.