Question

I’m writing a custom flex file to generate a lexer for use with JSyntaxpane.

The custom language I need to lex has different states that can be embedded into each other in a kind of stack.

I.E you could be writing an expression that has a single quoted string in it and then embed another expression within the string using a special token eval(). But you can also embed the expression within a double quoted string.

eg:

someExpressionFunction('a single-quoted string with an eval(expression) embedded in it', "a double-quoted string with an eval(expression) embedded in it")

This is a simplification, there are more states than this, but assuming I need to have different states for DOUBLE_STRING and SINGLE_STRING it adequately describes my situation.

What’s the best way to ensure I return to the correct state upon closing the eval expression (i.e return to DOUBLE_STRING if I was in double quotes, SINGLE_STRING if I was in single quotes)

The solution I’ve come up with, which works, is to keep track of state using a Stack and some custom methods to use in lieu of using yybegin to start a different state.

private Stack<Integer> stack = new Stack<Integer>();

public void yypushState(int newState) {
  stack.push(yystate());
  yybegin(newState);
}

public void yypopState() {
  yybegin(stack.pop());
}

Is this the best way to achieve this? Is there a simpler built-in function of JFlex I can leverage or a best practice I should know about?

Answer 1

I think that’s one very good way of doing it. I actually needed some similar feature to add Groovy GString, Python like String and some HTML to JavaDocs.

What I would also like to add is a Lexer calling a Lexer to parse sub sections. Something like JavaScript embedded in HTML. But I could not get the time to do it.

I like StackOverflow, but just wondering why didn’t you post this on JSyntaxPane’s issues?