Thread: Regular Expression Difficulties...

Hi everyone!

Right now I'm working on a regular expression that searches for comparison operators (">", "<=", "!=", etc.).

Here's the regex I've come up with:

private static final String C = "[\\Q><!\\E]?\\=?"; //comparator regex

It looks for a ">", "<", or "!" once or not at all, followed by a "=" once or not at all. The "\\Q" starts a quote (because "<", ">", and "!" are reserved characters for regular expressions, I need to escape to use them) that is ended by the "\\E". The "\\" in front of the "=" is again an escape, because "=" is also a reserved regex character. All of these have two "\" because the compiler throws an illegal escape character otherwise; the first "\" lets it know that the second "\" should be taken literally.

So, I whipped that up, thought it would work, and was sorely disappointed...

I tested it using this simple code:

public static void main(String[] teehee) {
  String s = "> < >= <= = !=";
  Matcher m;
  for(String part : s.split(" ")) {
    m = Pattern.compile(C).matcher(part); //C being the regex I described above
    System.out.print("\'" + part + "\': ");
    if(m.find()) {
      System.out.println("Valid!");
    } else {
      System.out.println("Invalid...");
    }
  }
}

Fortunately, when I ran that test, this was the result:

'>': Valid!
'<': Valid!
'>=': Valid!
'<=': Valid!
'=': Valid!
'!=': Valid!

Unforunately, this regex gets so carried away that it accepts almost anything...
Changing the value of 's' to ">as fqq<f >fq= <f=Q =D !s=", this results:

'>as': Valid!
'fqq<f': Valid!
'>fq=': Valid!
'<f=Q': Valid!
'=D': Valid!
'!s=': Valid!

...what have I done wrong? I'd appreciate any help!

Try adding some boundary matchers in there...word boundaries, or as the below shows, beginning and ending of the line

^[\\Q><!\\E]?\\=?$

Note the regex will also match an empty String (none of either)

Oh, alright! That makes perfect sense! All of those other wrong inputs did match, because they contained the characters I asked for. Never did I realize my assumption that the Matcher would only start at the beginning... Thanks very much!!

If I'm planning on putting this regex within other regex's (odd plural...), should I remove the "^" and "$"?

Thanks for the link; but I've already been reading that extensively.

I do the same with the pronunciation. Regex is so much easier to say than regular expression, though I sound awfully lazy saying that...