RegExp: lastIndex

The lastIndex data property of a RegExp instance specifies the index at which to start the next match.

Try it

Value

A non-negative integer.

Property attributes of RegExp: lastIndex
Writable yes
Enumerable no
Configurable no

Description

This property is set only if the regular expression instance used the g flag to indicate a global search, or the y flag to indicate a sticky search. The following rules apply when exec() is called on a given input:

  • If lastIndex is greater than the length of the input, exec() will not find a match, and lastIndex will be set to 0.
  • If lastIndex is equal to or less than the length of the input, exec() will attempt to match the input starting from lastIndex.
    • If exec() finds a match, then lastIndex will be set to the position of the end of the matched string in the input.
    • If exec() does not find a match, then lastIndex will be set to 0.

Other regex-related methods, such as RegExp.prototype.test(), String.prototype.match(), String.prototype.replace(), etc., call exec() under the hood, so they have different effects on lastIndex. See their respective pages for details.

Examples

Using lastIndex

Consider the following sequence of statements:

js

const re = /(hi)?/g;

Matches the empty string.

js

console.log(re.exec("hi"));
console.log(re.lastIndex);

Returns ["hi", "hi"] with lastIndex equal to 2.

js

console.log(re.exec("hi"));
console.log(re.lastIndex);

Returns ["", undefined], an empty array whose zeroth element is the match string. In this case, the empty string because lastIndex was 2 (and still is 2) and hi has length 2.

Using lastIndex with sticky regexes

The lastIndex property is writable. You can set it to make the regex start its next search at a given index.

The y flag almost always requires setting lastIndex. It always matches strictly at lastIndex and does not attempt any later positions. This is usually useful for writing parsers, when you want to match tokens only at the current position.

js

const stringPattern = /"[^"]*"/y;
const input = `const message = "Hello world";`;

stringPattern.lastIndex = 6;
console.log(stringPattern.exec(input)); // null

stringPattern.lastIndex = 16;
console.log(stringPattern.exec(input)); // ['"Hello world"']

Rewinding lastIndex

The g flag also benefits from setting lastIndex. One common use case is when the string is modified in the middle of a global search. In this case, we may miss a particular match if the string is shortened. We can avoid this by rewinding lastIndex.

js

const mdLinkPattern = /\[[^[\]]+\]\((?<link>[^()\s]+)\)/dg;

function resolveMDLink(line) {
  let match;
  let modifiedLine = line;
  while ((match = mdLinkPattern.exec(modifiedLine))) {
    const originalLink = match.groups.link;
    const resolvedLink = originalLink.replaceAll(/^files|\/index\.md$/g, "");
    modifiedLine =
      modifiedLine.slice(0, match.indices.groups.link[0]) +
      resolvedLink +
      modifiedLine.slice(match.indices.groups.link[1]);
    // Rewind the pattern to the end of the resolved link
    mdLinkPattern.lastIndex += resolvedLink.length - originalLink.length;
  }
  return modifiedLine;
}

console.log(
  resolveMDLink(
    "[`lastIndex`](files/en-us/web/javascript/reference/global_objects/regexp/lastindex/index.md)",
  ),
); // [`lastIndex`](/en-us/web/javascript/reference/global_objects/regexp/lastindex)
console.log(
  resolveMDLink(
    "[`ServiceWorker`](files/en-us/web/api/serviceworker/index.md) and [`SharedWorker`](files/en-us/web/api/sharedworker/index.md)",
  ),
); // [`ServiceWorker`](/en-us/web/api/serviceworker) and [`SharedWorker`](/en-us/web/api/sharedworker)

Try deleting the mdLinkPattern.lastIndex += resolvedLink.length - originalLink.length line and running the second example. You will find that the second link is not replaced correctly, because the lastIndex is already past the link's index after the string is shortened.

Warning: This example is for demonstration only. To deal with Markdown, you should probably use a parsing library instead of regex.

Optimizing searching

You can optimize searching by setting lastIndex to a point where previous possible occurrences can be ignored. For example, instead of this:

js

const stringPattern = /"[^"]*"/g;
const input = `const message = "Hello " + "world";`;

// Pretend we've already dealt with the previous parts of the string
let offset = 26;
const remainingInput = input.slice(offset);
const nextString = stringPattern.exec(remainingInput);
console.log(nextString[0]); // "world"
offset += nextString.index + nextString.length;

Consider this:

js

stringPattern.lastIndex = offset;
const nextString = stringPattern.exec(remainingInput);
console.log(nextString[0]); // "world"
offset = stringPattern.lastIndex;

This is potentially more performant because we avoid string slicing.

Avoiding side effects

The side effects caused by exec() can be confusing, especially if the input is different for each exec().

js

const re = /foo/g;
console.log(re.test("foo bar")); // true
console.log(re.test("foo baz")); // false, because lastIndex is non-zero

This is even more confusing when you are hand-modifying lastIndex. To contain the side effects, remember to reset lastIndex after each input is completely processed.

js

const re = /foo/g;
console.log(re.test("foo bar")); // true
re.lastIndex = 0;
console.log(re.test("foo baz")); // true

With some abstraction, you can require lastIndex to be set to a particular value before each exec() call.

js

function createMatcher(pattern) {
  // Create a copy, so that the original regex is never updated
  const regex = new RegExp(pattern, "g");
  return (input, offset) => {
    regex.lastIndex = offset;
    return regex.exec(input);
  };
}

const matchFoo = createMatcher(/foo/);
console.log(matchFoo("foo bar", 0)[0]); // "foo"
console.log(matchFoo("foo baz", 0)[0]); // "foo"

Specifications

Specification
ECMAScript Language Specification
# sec-properties-of-regexp-instances

Browser compatibility

BCD tables only load in the browser

See also