Question

I am writing a flat file parser that reads token/value pairs using a Scanner. The files being read contain the token “class”. The token is later used in a switch statement, and uses the (pre Java 7) valueOf(token) Java idiom to produce an enum value. (I am using Java6 for compatibility with GWT.) As a workaround, I am using uppercase values in the enum, and valueOf(token.toUpperCase()).

public enum ParseTags {
    CODE, CLASS, INSTRUCTOR, HOURS;
}

// . . . 

token = scanner.next();
value = scanner.next();

switch (ParseTags.valueOf(token.toUpperCase())) {
case CODE:
    entry.setCode(value);
    break;
case CLASS:
    entry.setClass(value);
    break;

Because this is being compiled into javascript, I want to avoid the extra “toUpperCase()” operation on each iteration; not sure what performance will be on target platform. Is there a more graceful way to represent reserved words in an enumeration? This would be handled well by Java7’s switch on String, but again, I am confined to Java6sdk.

Answer 1

What you’re doing right now is the preferred way to do it. I would be extraordinarily shocked if the toUpperCase were a bottleneck.

That said, I might consider something like

enum ParseTags {
  CODE {
    public void set(Entry entry, String value) { 
      entry.setCode(value);
    }
  },
  ...;
  public abstract void set(Entry entry, String value);
}

so you can do

ParseTags.valueOf(token.toUpperCase()).set(entry, value);