Is there a version of JavaScript's String.indexOf() that allows for regular expressions?

Viewed 205214

In javascript, is there an equivalent of String.indexOf() that takes a regular expression instead of a string for the first first parameter while still allowing a second parameter ?

I need to do something like

str.indexOf(/[abc]/ , i);

and

str.lastIndexOf(/[abc]/ , i);

While String.search() takes a regexp as a parameter it does not allow me to specify a second argument!

Edit:
This turned out to be harder than I originally thought so I wrote a small test function to test all the provided solutions... it assumes regexIndexOf and regexLastIndexOf have been added to the String object.

function test (str) {
    var i = str.length +2;
    while (i--) {
        if (str.indexOf('a',i) != str.regexIndexOf(/a/,i)) 
            alert (['failed regexIndexOf ' , str,i , str.indexOf('a',i) , str.regexIndexOf(/a/,i)]) ;
        if (str.lastIndexOf('a',i) != str.regexLastIndexOf(/a/,i) ) 
            alert (['failed regexLastIndexOf ' , str,i,str.lastIndexOf('a',i) , str.regexLastIndexOf(/a/,i)]) ;
    }
}

and I am testing as follow to make sure that at least for one character regexp, the result is the same as if we used indexOf

//Look for the a among the xes
test('xxx');
test('axx');
test('xax');
test('xxa');
test('axa');
test('xaa');
test('aax');
test('aaa');

22 Answers

Instances of the String constructor have a .search() method which accepts a RegExp and returns the index of the first match.

To start the search from a particular position (faking the second parameter of .indexOf()) you can slice off the first i characters:

str.slice(i).search(/re/)

But this will get the index in the shorter string (after the first part was sliced off) so you'll want to then add the length of the chopped off part (i) to the returned index if it wasn't -1. This will give you the index in the original string:

function regexIndexOf(text, re, i) {
    var indexInSuffix = text.slice(i).search(re);
    return indexInSuffix < 0 ? indexInSuffix : indexInSuffix + i;
}

Combining a few of the approaches already mentioned (the indexOf is obviously rather simple), I think these are the functions that will do the trick:

function regexIndexOf(string, regex, startpos) {
    var indexOf = string.substring(startpos || 0).search(regex);
    return (indexOf >= 0) ? (indexOf + (startpos || 0)) : indexOf;
}

function regexLastIndexOf(string, regex, startpos) {
    regex = (regex.global) ? regex : new RegExp(regex.source, "g" + (regex.ignoreCase ? "i" : "") + (regex.multiLine ? "m" : ""));
    if(typeof (startpos) == "undefined") {
        startpos = string.length;
    } else if(startpos < 0) {
        startpos = 0;
    }
    var stringToWorkWith = string.substring(0, startpos + 1);
    var lastIndexOf = -1;
    var nextStop = 0;
    while((result = regex.exec(stringToWorkWith)) != null) {
        lastIndexOf = result.index;
        regex.lastIndex = ++nextStop;
    }
    return lastIndexOf;
}

UPDATE: Edited regexLastIndexOf() so that is seems to mimic lastIndexOf() now. Please let me know if it still fails and under what circumstances.


UPDATE: Passes all tests found on in comments on this page, and my own. Of course, that doesn't mean it's bulletproof. Any feedback appreciated.

You could use substr.

str.substr(i).match(/[abc]/);

Based on BaileyP's answer. The main difference is that these methods return -1 if the pattern can't be matched.

Edit: Thanks to Jason Bunting's answer I got an idea. Why not modify the .lastIndex property of the regex? Though this will only work for patterns with the global flag (/g).

Edit: Updated to pass the test-cases.

String.prototype.regexIndexOf = function(re, startPos) {
    startPos = startPos || 0;

    if (!re.global) {
        var flags = "g" + (re.multiline?"m":"") + (re.ignoreCase?"i":"");
        re = new RegExp(re.source, flags);
    }

    re.lastIndex = startPos;
    var match = re.exec(this);

    if (match) return match.index;
    else return -1;
}

String.prototype.regexLastIndexOf = function(re, startPos) {
    startPos = startPos === undefined ? this.length : startPos;

    if (!re.global) {
        var flags = "g" + (re.multiline?"m":"") + (re.ignoreCase?"i":"");
        re = new RegExp(re.source, flags);
    }

    var lastSuccess = -1;
    for (var pos = 0; pos <= startPos; pos++) {
        re.lastIndex = pos;

        var match = re.exec(this);
        if (!match) break;

        pos = match.index;
        if (pos <= startPos) lastSuccess = pos;
    }

    return lastSuccess;
}

RexExp instances have a lastIndex property already (if they are global) and so what I'm doing is copying the regular expression, modifying it slightly to suit our purposes, exec-ing it on the string and looking at the lastIndex. This will inevitably be faster than looping on the string. (You have enough examples of how to put this onto the string prototype, right?)

function reIndexOf(reIn, str, startIndex) {
    var re = new RegExp(reIn.source, 'g' + (reIn.ignoreCase ? 'i' : '') + (reIn.multiLine ? 'm' : ''));
    re.lastIndex = startIndex || 0;
    var res = re.exec(str);
    if(!res) return -1;
    return re.lastIndex - res[0].length;
};

function reLastIndexOf(reIn, str, startIndex) {
    var src = /\$$/.test(reIn.source) && !/\\\$$/.test(reIn.source) ? reIn.source : reIn.source + '(?![\\S\\s]*' + reIn.source + ')';
    var re = new RegExp(src, 'g' + (reIn.ignoreCase ? 'i' : '') + (reIn.multiLine ? 'm' : ''));
    re.lastIndex = startIndex || 0;
    var res = re.exec(str);
    if(!res) return -1;
    return re.lastIndex - res[0].length;
};

reIndexOf(/[abc]/, "tommy can eat");  // Returns 6
reIndexOf(/[abc]/, "tommy can eat", 8);  // Returns 11
reLastIndexOf(/[abc]/, "tommy can eat"); // Returns 11

You could also prototype the functions onto the RegExp object:

RegExp.prototype.indexOf = function(str, startIndex) {
    var re = new RegExp(this.source, 'g' + (this.ignoreCase ? 'i' : '') + (this.multiLine ? 'm' : ''));
    re.lastIndex = startIndex || 0;
    var res = re.exec(str);
    if(!res) return -1;
    return re.lastIndex - res[0].length;
};

RegExp.prototype.lastIndexOf = function(str, startIndex) {
    var src = /\$$/.test(this.source) && !/\\\$$/.test(this.source) ? this.source : this.source + '(?![\\S\\s]*' + this.source + ')';
    var re = new RegExp(src, 'g' + (this.ignoreCase ? 'i' : '') + (this.multiLine ? 'm' : ''));
    re.lastIndex = startIndex || 0;
    var res = re.exec(str);
    if(!res) return -1;
    return re.lastIndex - res[0].length;
};


/[abc]/.indexOf("tommy can eat");  // Returns 6
/[abc]/.indexOf("tommy can eat", 8);  // Returns 11
/[abc]/.lastIndexOf("tommy can eat"); // Returns 11

A quick explanation of how I am modifying the RegExp: For indexOf I just have to ensure that the global flag is set. For lastIndexOf of I am using a negative look-ahead to find the last occurrence unless the RegExp was already matching at the end of the string.

It does not natively, but you certainly can add this functionality

<script type="text/javascript">

String.prototype.regexIndexOf = function( pattern, startIndex )
{
    startIndex = startIndex || 0;
    var searchResult = this.substr( startIndex ).search( pattern );
    return ( -1 === searchResult ) ? -1 : searchResult + startIndex;
}

String.prototype.regexLastIndexOf = function( pattern, startIndex )
{
    startIndex = startIndex === undefined ? this.length : startIndex;
    var searchResult = this.substr( 0, startIndex ).reverse().regexIndexOf( pattern, 0 );
    return ( -1 === searchResult ) ? -1 : this.length - ++searchResult;
}

String.prototype.reverse = function()
{
    return this.split('').reverse().join('');
}

// Indexes 0123456789
var str = 'caabbccdda';

alert( [
        str.regexIndexOf( /[cd]/, 4 )
    ,   str.regexLastIndexOf( /[cd]/, 4 )
    ,   str.regexIndexOf( /[yz]/, 4 )
    ,   str.regexLastIndexOf( /[yz]/, 4 )
    ,   str.lastIndexOf( 'd', 4 )
    ,   str.regexLastIndexOf( /d/, 4 )
    ,   str.lastIndexOf( 'd' )
    ,   str.regexLastIndexOf( /d/ )
    ]
);

</script>

I didn't fully test these methods, but they seem to work so far.

After having all the proposed solutions fail my tests one way or the other, (edit: some were updated to pass the tests after I wrote this) I found the mozilla implementation for Array.indexOf and Array.lastIndexOf

I used those to implement my version of String.prototype.regexIndexOf and String.prototype.regexLastIndexOf as follows:

String.prototype.regexIndexOf = function(elt /*, from*/)
  {
    var arr = this.split('');
    var len = arr.length;

    var from = Number(arguments[1]) || 0;
    from = (from < 0) ? Math.ceil(from) : Math.floor(from);
    if (from < 0)
      from += len;

    for (; from < len; from++) {
      if (from in arr && elt.exec(arr[from]) ) 
        return from;
    }
    return -1;
};

String.prototype.regexLastIndexOf = function(elt /*, from*/)
  {
    var arr = this.split('');
    var len = arr.length;

    var from = Number(arguments[1]);
    if (isNaN(from)) {
      from = len - 1;
    } else {
      from = (from < 0) ? Math.ceil(from) : Math.floor(from);
      if (from < 0)
        from += len;
      else if (from >= len)
        from = len - 1;
    }

    for (; from > -1; from--) {
      if (from in arr && elt.exec(arr[from]) )
        return from;
    }
    return -1;
  };

They seem to pass the test functions I provided in the question.

Obviously they only work if the regular expression matches one character but that is enough for my purpose since I will be using it for things like ( [abc] , \s , \W , \D )

I will keep monitoring the question in case someone provides a better/faster/cleaner/more generic implementation that works on any regular expression.

For a solution that's more concise than most of the other answers posted, you may want to use the String.prototype.replace function which will run a function on every detected pattern. For example:

let firstIndex = -1;
"the 1st numb3r".replace(/\d/,(p,i) => { firstIndex = i; });
// firstIndex === 4

This is especially useful for the "last index" case:

let lastIndex = -1;
"the l4st numb3r".replace(/\d/g,(p,i) => { lastIndex = i; });
// lastIndex === 13

Here, it's important to include the "g" modifier so that all occurrences are evaluated. These versions will also result in -1 if the regular expression was not found.

Finally, here are the more general functions which include a start index:

function indexOfRegex(str,regex,start = 0) {
    regex = regex.global ? regex : new RegExp(regex.source,regex.flags + "g");
    let index = -1;
    str.replace(regex,function() {
        const pos = arguments[arguments.length - 2];
        if(index < 0 && pos >= start)
            index = pos;
    });
    return index;
}

function lastIndexOfRegex(str,regex,start = str.length - 1) {
    regex = regex.global ? regex : new RegExp(regex.source,regex.flags + "g");
    let index = -1;
    str.replace(regex,function() {
        const pos = arguments[arguments.length - 2];
        if(pos <= start)
            index = pos;
    });
    return index;
}

These functions specifically avoid splitting the string at the start index which I feel is risky in the age of Unicode. They don't modify the prototype of common Javascript classes (although you're free to do so yourself). They accept more RegExp flags, for example "u" or "s" and any flags that may be added in the future. And I find it easier to reason about callback functions than for/while loops.

The regexIndexOf from Jason Bunting can be inverted more simply and still support UTF8 characters by doing this:

function regexLastIndexOf(string, regex, startpos=0) {
    return text.length - regexIndexOf([...text].reverse().join(""), regex, startpos) - 1;
}

If you are looking for a very simple lastIndex lookup with RegExp and don't care if it mimics lastIndexOf to the last detail, this may catch your attention.

I simply reverse the string, and subtract the first occurence index from length - 1. It happens to pass my test, but I think there could arise a performance issue with long strings.

interface String {
  reverse(): string;
  lastIndex(regex: RegExp): number;
}

String.prototype.reverse = function(this: string) {
  return this.split("")
    .reverse()
    .join("");
};

String.prototype.lastIndex = function(this: string, regex: RegExp) {
  const exec = regex.exec(this.reverse());
  return exec === null ? -1 : this.length - 1 - exec.index;
};

I used String.prototype.match(regex) which returns a string array of all found matches of the given regex in the string (more info see here):

function getLastIndex(text, regex, limit = text.length) {
  const matches = text.match(regex);

  // no matches found
  if (!matches) {
    return -1;
  }

  // matches found but first index greater than limit
  if (text.indexOf(matches[0] + matches[0].length) > limit) {
    return -1;
  }

  // reduce index until smaller than limit
  let i = matches.length - 1;
  let index = text.lastIndexOf(matches[i]);
  while (index > limit && i >= 0) {
    i--;
    index = text.lastIndexOf(matches[i]);
  }
  return index > limit ? -1 : index;
}

// expect -1 as first index === 14
console.log(getLastIndex('First Sentence. Last Sentence. Unfinished', /\. /g, 10));

// expect 29
console.log(getLastIndex('First Sentence. Last Sentence. Unfinished', /\. /g));

var mystring = "abc ab a";
var re  = new RegExp("ab"); // any regex here

if ( re.exec(mystring) != null ){ 
   alert("matches"); // true in this case
}

Use standard regular expressions:

var re  = new RegExp("^ab");  // At front
var re  = new RegExp("ab$");  // At end
var re  = new RegExp("ab(c|d)");  // abc or abd

let regExp; // your RegExp here
arr.map(x => !!x.toString().match(regExp)).indexOf(true)

You can use String.prototype.matchAll(), together with the convenient Array.prototype.at():

const str = "foo a foo B";
const matches = [...str.matchAll(/[abc]/gi)];

if (matches.length) {
  const indexOfFirstMatch = matches.at(0).index;
  const indexOfLastMatch = matches.at(-1).index;
  console.log(indexOfFirstMatch, indexOfLastMatch)
}

Well, as you are just looking to match the position of a character , regex is possibly overkill.

I presume all you want is, instead of "find first of these this character" , just find first of these characters.

This of course is the simple answer, but does what your question sets out to do, albeit without the regex part ( because you didn't clarify why specifically it had to be a regex )

function mIndexOf( str , chars, offset )
{
   var first  = -1; 
   for( var i = 0; i < chars.length;  i++ )
   {
      var p = str.indexOf( chars[i] , offset ); 
      if( p < first || first === -1 )
      {
           first = p;
      }
   }
   return first; 
}
String.prototype.mIndexOf = function( chars, offset )
{
   return mIndexOf( this, chars, offset ); # I'm really averse to monkey patching.  
};
mIndexOf( "hello world", ['a','o','w'], 0 );
>> 4 
mIndexOf( "hello world", ['a'], 0 );
>> -1 
mIndexOf( "hello world", ['a','o','w'], 4 );
>> 4
mIndexOf( "hello world", ['a','o','w'], 5 );
>> 6
mIndexOf( "hello world", ['a','o','w'], 7 );
>> -1 
mIndexOf( "hello world", ['a','o','w','d'], 7 );
>> 10
mIndexOf( "hello world", ['a','o','w','d'], 10 );
>> 10
mIndexOf( "hello world", ['a','o','w','d'], 11 );
>> -1
Related