How to match overlapping substrings as whole words (with multiple occurence) with regex?

Question

I tried to follow this answer but it doesn't work when there are multiple occurrences of the same word.

I want to count the occurrence of both site and site web in the string "site web site".

I tried with the following code :

var regex = /(?:\b)((?=(site))(?=(site web)))(?:\b)/;
var string = 'site web site';
var match = string.match( regex ).filter(Boolean);

console.log(match)

This code returns ["site", "site web"] but I want it to return ["site", "site", "site web"] since site appears two times in the string.

Note : In my case, I have hundreds of words to match.

More, if the input is site webS site, the expected output is ["site", "site"]. The input is supposed to be a complete text with punctuations to take into account (.,?!/;...).

@WiktorStribiżew Thank you, but I don't want to alter my list of keywords since I'll probably have to match hundreds of keywords in a text of thousands of words. — Valentin Duboscq, Aug 08 '19 at 16:04
Not a problem, use `site|web site` regex, see https://jsfiddle.net/y49kq78v/ — Wiktor Stribiżew, Aug 08 '19 at 16:10
It's working but when I try with, let's say `site webS site`, it's still returning the same array `["site", "site web", "site"]` when it's not supposed to. It should return `["site", "site"]`. — Valentin Duboscq, Aug 08 '19 at 16:21

score 0 · Answer 1 · answered Aug 08 '19 at 20:44

If you need to find if the words of a list of words appears in a string and you have hundred of words, you need to use a good string searching algorithm. I think for your use case the best option is Aho Corasick algorithm. It achieves near to O(n) complexity which is much faster than using regexp.

Check this link:

Aho Corasick

I used it and I can say you it works very fast. If you decide to use it, there are available in github multiple implementations of this algorithm. You can search a good one for your programming language and use it.

I hope this helps.

How to match overlapping substrings as whole words (with multiple occurence) with regex?

1 Answers1