no-useless-escape rule is not handling regex back references correctly

[jvswatson]

It appears that the recent changes related to the no-useless-escape rule are not handling the integer back references (e.g. \1) correctly.

eslint: 3.9.0
node: 4.4.3
npm: 2.15.1
parser: default

.eslintrc:

{
  "rules": {
    "no-useless-escape": 2,
    "no-console": 0
  }
}

JavaScript file:

var keyword = "'the thing \"on the inside\" is what to match'";
var exactMatchPhrases = keyword.match(/(["'])(.*?)\1/g);
console.log(exactMatchPhrases);

What did you expect to happen?
No eslint errors

What actually happened? Please include the actual, raw output from ESLint.
2:51 error Unnecessary escape character: \1 no-useless-escape

[not-an-aardvark]

Confirmed, thanks for the report.

[not-an-aardvark]

I think the issue is that digits in regex literals should always be valid escapes (either they represent control characters, or they represent backreferences).

var foo = /\0/; // null character
var foo = /\1/; // \x01 character
var foo = /(a)\1/; // 2 'a' characters in a row

[ljharb]

Another example: /^(['"]).*\1$/