no-misleading-character-class

A character class such as [ab] matches one code point from the list. Many things that look like one character are several: a letter followed by a combining accent, an emoji with a skin tone, a flag, a family joined with zero-width joiners. Without the u flag, most emoji count as two. Inside [...] each part becomes its own alternative, so /^[👍]$/.test("👍") is false.

The rule checks regular expression literals and the patterns given to RegExp(...) and new RegExp(...).

--fix-suggestions adds the u flag when that's all that's missing and the pattern stays valid with it.

This rule is on by default.

Configure

On by default as an error. To turn it off:

[lint.rules]
"no-misleading-character-class" = "off" # or "warn"

Every option, set to its default:

[lint.rules]
"no-misleading-character-class" = ["error", {
  allowEscape = false,
}]

Examples

Reported:

const isThumbsUp = /^[👍]$/;

const isFlag = /^[🇯🇵🇩🇪]$/u;

const isFamily = new RegExp("[👨‍👩‍👦]", "u");

Not reported:

const isThumbsUp = /^[👍]$/u;

const isFlag = /^(?:🇯🇵|🇩🇪)$/u;

Options

Option	Default	What it does
allowEscape	false	Don't report code points that are written as escapes, such as \u0301. The escape shows that you meant a separate code point.

const accents = /[A\u0301]/;

const accents = /[A\u0301]/;