Handling Antlr Syntax Errors or how to give a better message on unexpected token

Question

We have the following sub-part of an Antlr grammar:

signed_int
        : SIGN? INT
    ;

INT : '0'..'9'+
        ;

When someone enters a numeric value everything is fine, but if they mistakenly type something like 1O (one and capital o) we get a cryptic error message like:

error 1 : Missing token  at offset 14
near [Index: 0 (Start: 0-Stop: 0) ='<missing COLON>'     type<24> Line: 26 LinePos:14]
 : syntax error...

What is a good way to handle this type of error? I thought of defining catch-all SYMBOL token type but this lead to too many parser building errors. I will continue looking into Antlr error handling but I thought I would post this here to look for some insights.

Answer 1

You should Override the reportError methods in lexer and parser. You can do it by adding this code to your lexer file:

  @Override
public void reportError(RecognitionException e) {
    throw new RuntimeException(e);
}

And create a method matches in parser that checks if input string matches the specified grammar:

 public static boolean matches(String input) {
     try {
         regExLexer lexer = new regExLexer(new ANTLRStringStream(input));
         regExParser parser = new regExParser(new CommonTokenStream(lexer));
         parser.goal();
         return true;
     } catch (RuntimeException e) {
         return false;
     }
     catch (Exception e) {
         return false;
     }
     catch (OutOfMemoryError e) {
         return false;
     }

 }

 @Override
 public void reportError(RecognitionException e) {
     throw new RuntimeException(e);
 }

Then in your file use the Parser.matches(input); to check if the given input matches the gramar. If it matches the method returns true, otherwise returns false, so when it returns false you can give any customized error message to users.