Question

It is common knowledge that the exec method from RegExps can get used to find all matches in a string. However, I just found out that if the regexp matches the empty string this loop can get stuck forever

var s = '1234'
var r = /()/g;
var m;
var i = 0;
while( (m = r.exec(s)) ){
   console.log(i, m[0]);
   if(++i >= 50){ console.log("infinite loop!"); break }
}

What is really weird though is that the plain string.match methods does not get stuck:

'1234'.match(/()/g) // Gives ["", "", "", "", "", ""]

I wonder how the match method is defined to work differently from the exec loop. So far the only way I found to avoid getting stuck like the match method does involves abusing the string.replace method, in a horrid hack:

var matches = [];
'1234'.replace(/()/g, function(m){ matches.push(m) });

So my question is:

How do the match and replace return a finite results when the regexp matches the empty string? Can I use the same technique to avoid getting stuck in the exec loop?

Answer 1

One (icky) solution is to make sure it’s not in the same place as last time:

var s = '1234'
var r = /()/g;
var m;
var i = 0;
var lastPosition = -1;

while(m = r.exec(s)) {
    if(m.index === lastPosition) {
        r.lastIndex++;
        continue;
    }

    lastPosition = m.index;

    console.log(i, m[0]);
}