How to use OR in Regular Expressions?

[Peter Heide]

I need to write a regular expression according to the following rules:

String starts with an "A"
String continues with 4 to 6 uppercase letters "A-Z" or digits "0-9"

OR

String starts with an uppercase letter from "B-Z"
String continues with 6 digits "0-9"

The following code uses two regular expressions, one for the first case and one for the second case.

It works, but it would be better and shorter to put everything into one single regular expression. I believe it is possible because regular expressions are very powerful, but I do not know how...

Has anyone an idea how to do this?

package de.allianz.bid_generator;
import java.util.regex.Matcher;
import java.util.regex.Pattern;
 
public class RegularExpression {
public static void validateID(String id) {
Pattern p1 = Pattern.compile("[A]{1}[0-9]{6}");
Matcher m1 = p1.matcher(id);
boolean res1 = m1.matches();
Pattern p2 = Pattern.compile("[B-Z]{1}[A-Z0-9]{4,6}");
Matcher m2 = p2.matcher(id);
boolean res2 = m2.matches();
boolean res = res1 | res2;
System.out.println("ID " + id + " is " + res); 
}

public static void main(String[] args) {
validateID("A123456"); // true
validateID("AB12345"); // false
validateID("B1234"); // true
validateID("BBBBBBBBBBBB"); // false
}
}

[ April 14, 2008: Message edited by: Peter Heide ]

[Campbell Ritchie]

You can use the | operator. I suggest a few hours reading the Java Tutorial about regular expressions will help no end.

BTW: Please write "boolean res = res1 || res2;" rather than using |.

[Peter Heide]

Thank you Richie, your proposal works fine using the OR (|) is the regular expression:

    
"[A]{1}[0-9]{6}|[B-Z]{1}[A-Z0-9]{4,6}"

Sometimes when you have a problem it looks quite complicated. When you see the solution it looks so easy.

[Jim Yingst]

[Campbell]: BTW: Please write "boolean res = res1 || res2;" rather than using |.

I agree with this advice, but just to clarify: you should generally prefer || to | for an OR in boolean operations, e.g. in conditional expressions. This is not to be confused with the OR inside a regex, where | is what you need. Also in this particular example there is no real difference between | and ||, however in many other applications there is a difference. Generally you only want | if you're using integers, not booleans, and you want to perform bitwise operations like

int a = 0x5C;    // binary 01011100
int b = 0x0F;    // binary 00001111
int c = a | b;   // binary 01011111

If you want to operate on each bit individually like this, then | is what you want. But for boolean expressions with a single boolean result, || is clearer and also has the advantage of short-circuiting - not evaluating the second operand if it's not necessary. This is what you want probably 99% of the time, so it's good to get in the habit of writing || rather than |.

The same arguments would also apply to writing && rather than & for AND.

[Campbell Ritchie]

Originally posted by Peter Heide:
Thank you Richie.

You're welcome. And thank you, Jim for the explanation of || which I ought probably to have given myself.

[Guido Sautter]

Just a thought ... using | in boolean expressions is sometimes quite essential, especially when working with the Collections Framework. Think of this:

Object a = new Object();
Object b = new Object();
 
Set s = new HashSet();
Set t = new HashSet();

Then the outcome of

boolean k = (s.add(a) || t.add(b));

is quite different from the outcome of

boolean k = (s.add(a) | t.add(b));

basicallly because in the first way, t.add(b) is never touched unless Set s already contains Object a.

[Jim Yingst]

Guido: true. I rarely see code that uses | like that, and personally I would probably rewrite it (or insert a comment) in order to make it more obvious that using | wasn't just a typo. Just because people aren't used to seeing it, and someone may try to "fix" it later without realizing the true intent.

[Guido Sautter]

Hi Jim,

in productive code I'd surely add some comment. Plus, programming with such side effects isn't quite the style of code that is easy to read. Whole thing was just a thought ...