regex - Javascript replace method, replace with "$1"

Question

I'm reading Sitepoints 2007 book "Simply Javascript" and I encountered some code I just can't understand.

It's the following code:

Core.removeClass = function(target, theClass)
{
    var pattern = new RegExp("(^| )" + theClass + "( |$)");
    target.className = target.className.replace(pattern, "$1");
    target.className = target.className.replace(/ $/, "");
};

The first call to the replace method is what puzzles me, I don't understand where the "$1" value comes from or what it means. I would think that the call should replace the found pattern with "".

Answer 1

Each pair of parentheses (...) where the first character is not a ?* is a "capturing group", which places its result into $1,$2,$3,etc which can be used in the replacement pattern.

You might also see the same thing as 1,2,3 in other regex engines, (or indeed in the original expression sometimes, for repetition)

These are called "backreferences", because they generally refer back to (an earlier) part of in the expression.

(*The ? indicates various forms of special behaviour, including a non-capturing group which is (?:...) and simply groups without capturing.)

In your specific example, the $1 will be the group (^| ) which is "position of the start of string (zero-width), or a single space character".

So by replacing the whole expression with that, you're basically removing the variable theClass and potentially a space after it. (The closing expression ( |$) is the inverse - a space or the string end position - and since its value isn't used, could have been non-capturing with (?: |$) instead.)

Hopefully this explains everything ok - let me know if you want any more info.

Also, here's some further reading from the site regular-expressions.info:

• Groups and Backreferences
• Atomic Grouping (doesn't work in JS, but interesting)
• Lookaround groups (partial support in JS regex)