Warm tip: This article is reproduced from stackoverflow.com, please click
regex

Zero-Length regexes and infinite matches?

发布于 2020-03-28 23:13:40

In trying to elaborate an answer to this question, I am now trying to come to terms with the behavior/meaning of Zero-Length regular expressions.

I often use www.regexr.com as a playground to test/debug/understand what's going on in regular expressions.

So we have this most banal scenario:

The regex is a*

The input string is dgwawa (As a matter of fact, the string here is irrelevant)

Why this behavior of reporting that this regex will match infinitely, since it matches zero occurrences of the preceding character ?

Why can't the result be 6 matches, one for each character position (since at every character, regardless of whether it is an a or not, there is a match, since zero matches is a match)?

How does it get into matching infinitely ? So it does not check/progress a character at a time?

I wonder how/where does it get itself into an infinite loop.

enter image description here

Questioner
Veverke
Viewed
72
Wiktor Stribiżew 2020-01-31 17:33

You selected JavaScript regex flavor at regexr.com online regex tester. JavaScript regex engine does not move the index automatically when a pattern that can match an empty string is passed.

That is why when you need to emulate the behavior observed in .NET Regex.Matches, PHP preg_match_all, Python re.finditer, etc. you need to manually advance the index to test each position.

See regex101.com test:

var re = /a*/g; 
var str = 'dgwawa';
var m;
 
while ((m = re.exec(str)) !== null) {
    if (m.index === re.lastIndex) {   // <- this part
        re.lastIndex++;               // <- here
    }                                 // <- is important
    document.body.innerHTML += "'" + m[0] + "'<br/>";
}

If you remove that if block, you will get an infinite loop.

There are two very important things to mention with this regard:

  • Always use appropriate online regex tester for your programming language
  • Avoid using unanchored patterns that can match empty strings