#no-misleading-character-class

#Configuration

import { defineConfig, js } from '@rslint/core';

export default defineConfig([
  js.configs.recommended,
  {
    rules: {
      'no-misleading-character-class': 'error',
    },
  },
]);

#Rule Details

Disallow characters whose visual rendering is made from multiple code points (combining marks, surrogate pairs, regional indicators, emoji-modifier sequences, joined ZWJ sequences) from appearing inside a regex character class [...]. Such sequences cannot be matched as a single unit by the regex engine and therefore produce surprising matches.

Examples of incorrect code for this rule:

/^[Á]$/u;        // a + combining acute
/^[❇️]$/u;       // base + variation selector
/^[👶🏻]$/u;     // base emoji + skin tone modifier
/^[🇯🇵]$/u;     // two regional indicator symbols
/^[👨‍👩‍👦]$/u; // ZWJ-joined family sequence
/^[👍]$/;        // astral character without `u` / `v` flag
new RegExp("[🎵]");

Examples of correct code for this rule:

/^[abc]$/;
/^[👍]$/u;
/^[\q{👶🏻}]$/v; // v-flag grouping preserves the sequence
new RegExp("^[]$");
/[\ud83d\udc4d]/;
/[\u00B7\u0300-\u036F]/u;

#Options

This rule accepts an options object:

{ "no-misleading-character-class": ["error", { "allowEscape": true }] }

• allowEscape (boolean, default false) — when enabled, allows combining the troublesome characters inside a character class as long as the combining portion is written using a backslash escape sequence. This applies to regex literals and to RegExp(...) calls whose first argument is a string or no-substitution template literal.

Examples of correct code with { "allowEscape": true }:

/[\ud83d\udc4d]/;                // surrogate pair written with escapes
/[A\u0301]/;                     // combining acute written with escape
new RegExp("[\\uD83D\\uDC4D]");  // surrogate pair in string literal

#Differences from ESLint

• The scope of recognition for the RegExp(...) constructor is limited to calls where the callee is a recognized global RegExp constructor and the pattern or flags are statically known strings. Spread arguments and unresolved runtime expressions are ignored.
• When a regex literal is passed to RegExp(...) with an unresolved second flags argument, rslint does not report on the literal because the runtime flags may override the literal's own flags.
• Suggestions to add the u flag use a simplified heuristic to decide whether the pattern remains valid under the flag. Patterns with identity escapes on letters (e.g. /[👍]\a/) are correctly detected as unfixable; more exotic cases may rarely produce a suggestion that ESLint would suppress.
• In rare edge cases where a character written as \q{...} or \p{...} escape appears outside its valid flag context (e.g. without the v flag), the scanner may still treat it as a breaker rather than a plain character.

#Original Documentation

• https://eslint.org/docs/latest/rules/no-misleading-character-class