Trying to find indices of all regex matches, but some being left out

Question

I want to find the index of every vowel after the first "e" in a string.

Since you can't get the index of the capture group directly from RegExp.exec(sInput), but you can get the length of a capture group containing everything in front of the actual capture group, the regex I'm using to do this is /(.*?e.*?)(a|e|i|o|u)(.*)/.

So the setup is basically like this:

let re = /(.*?e.*?)(a|e|i|o|u)(.*)/g;
let sInput = "lorem ipsum";

let tMatches = [];
let tMatchIndices = [];
let iPrevIndex = 0;

while (result = re.exec(sInput)) {
    /*  result[0]: full match
        result[1]: match for 1st capture group (.*?e.*?)
        result[2]: match for 2nd capture group (a|e|i|o|u)
        result[3]: match for 3rd capture group (.*)
    */
    let index = result[1].length + iPrevIndex;
    let sMatch = result[2];
    tMatchIndices.push(index);
    tMatches[index] = sMatch;
    iPrevIndex = index + sMatch.length;
    re.lastIndex = iPrevIndex;
}

for (i = 0; i < tMatches.length; i++) {
  let index = tMatchIndices[i];
    console.log(tMatches[index] + " at index "+index);
}

The issue is for the input string "lorem ipsum", I need the indices for both the "i" and the "u"... and it's only giving me the index for "i".

I know why it's doing this - advancing the search index past the first match cuts out the "e" that's supposed to trigger the next match. What I'm stuck on is how to fix it. I can't just simply not advance the search index, or it would never move past the first match anyway.

I've thought about simply deleting each match from the search string as I go along, but then that shifts the index of every character after it to the left, so the indices I would collect wouldn't even be accurate for the original, untruncated string.

What do?

user_1863229 · Answer

One approach to keep things simple would be to strip off the leading substring up to, and including, the first e. Then, just iterate the remaining string one character at a time, checking for vowels along the way.

sInput = "lorem ipsum";
nInput = sInput.replace(/^.*?(?:e|$)/, "");
var index = sInput.length - nInput.length;
var indices = [];
var counter = 0;
for (var i=0; i < nInput.length; i++) {
    if (/[aeiou]/.test(nInput.charAt(i))) {
        indices[counter++] = i + index;
    }
}

console.log(indices);

Regarding the output:

01234567890
lorem ipsum
      ^  ^  [6, 9]

user_7475450 · Answer

You can do that with a positive lookbehind:

'lorem ipsum'.replace(/(?<=e.*)[aiueo]/g, function(m, offset) {
  console.log(m + ' ==> ' + offset)
});

Output:

i ==> 6
u ==> 9

Explanation:

(?<=e.*) - positive lookbehind for character e
[aiueo] - scan for a vowel
use the g flag to repeat
in the replace function you can reference the offset

user_19416209 · Answer

Using the String.prototype.matchAll, which returns an iterator of matches, each being an array with the index property:

[...str.matchAll(/(?<=^.*?e.*?)[aeiou]/gis)].map(m => m.index);

Demo:

const findIndices = str => {
  const regex = /(?<=^.*?e.*?)[aeiou]/gis;
  return [...str.matchAll(regex)].map(m => m.index);
}

console.log(findIndices('Lorem ipsum dolor sit amet.'));

.as-console-row-code { white-space: normal !important; }

Trying to find indices of all regex matches, but some being left out

3 Answers