CODEWITHSUNDEEP
regexparsinginputantlr
d m
d m

Reputation: 21

ANTLR match to end of input

I'm trying to match my grammar to an entire string, and have it error out if it cannot consume the entire input. Basically, this pseudo regex:

\whitespace*  [a-zA-Z]+  [-]?  [a-zA-Z]+ \whitespace* $

According to this, EOF should work. So, consider this grammar:

start : CHARS EOF
      ;

CHARS : ('a'..'z')+
      ;

If I set input to "hello" in this code:

        var ss = new Antlr.Runtime.ANTLRStringStream(input);
        var lex = new autocompleteLexer(ss);
        var tokens = new CommonTokenStream(lex);
        var parser = new autocompleteParser(tokens);
        return (BaseTree)(parser.start().Tree);

I get an AST with two children, HELLO and EOF, as expected. But if I set the input to say "hello#", it gives back the same tree, and doesn't complain about the hash at all. I'm not sure what I'm doing wrong.

Upvotes: 2

Views: 1416

Answers (1)

d m
d m

Reputation: 21

I found the problem, it was that exceptions were getting trapped by the Lexer instead of propagating into my code. So, by adding this:

class ThrowAllLexer: queryLexer
{
    public ThrowAllLexer(ICharStream input): base(input) { }
    public override void  ReportError(RecognitionException e) {
        throw e;
}

I now get the proper exceptions, instead of having them be swallowed.

Upvotes: 2

Related Questions