Strings and regular expressions: lengths, flags and the stateful g

A string is an immutable sequence of UTF-16 code units. Every method — toUpperCase, slice, replace, trim — returns a new string. Assigning to s[0] never changes the string: in non-strict code it is silently ignored, in strict code it throws a TypeError. A reassigned variable gets a new string; the old one is untouched.

length counts code units, not characters

Characters outside the Basic Multilingual Plane, such as most emoji, take two code units, so "👍".length is 2. Spreading a string or using for...of iterates by code point instead. Accented letters can also be written two ways: é as one code point, or e followed by a combining accent. They look identical, but they are different strings until you call normalize() on both.

1export function run() {2  const composed = "caf\u00e9";       // é as one code point3  const decomposed = "cafe\u0301";    // e + combining acute accent4  const thumbs = "\u{1F44D}";         // a single emoji5 6  let s = "abc";7  s[0] = "X";                          // ignored: strings are immutable8  let strictWrite;9  try {10    (() => {11      "use strict";12      s[0] = "X";13    })();14  } catch (e) {15    strictWrite = e.name;16  }17 18  return {19    composedLength: composed.length,20    decomposedLength: decomposed.length,21    sameString: composed === decomposed,22    sameAfterNormalize: composed.normalize() === decomposed.normalize(),23    emojiLength: thumbs.length,24    emojiCodePoints: [...thumbs].length,25    afterIndexAssign: s,26    strictWrite,27    replaceFirst: "a-b-c".replace("-", "+"),28    replaceEvery: "a-b-c".replaceAll("-", "+"),29    lastChar: "hello".at(-1),30    padded: "7".padStart(3, "0"),31  };32}

The everyday methods: slice(start, end) (negative indexes count from the end), includes, startsWith, endsWith, padStart, split/join, trim, at(-1). replace with a string pattern replaces only the first match; replaceAll, or a regex with the g flag, replaces every one.

Regular expressions

A regex literal is /pattern/flags; new RegExp(source, flags) builds one from a string (where every backslash must itself be escaped). The flags you will meet: g (all matches), i (ignore case), m (^ and $ match at line breaks), s (. also matches newlines), u (Unicode code points), (sticky: match only at ).

1export function run() {2  const isCat = /cat/g;3  const sharedGlobal = [isCat.test("cat"), isCat.test("cat"), isCat.test("cat")];4 5  const html = "<b>bold</b>";6  const date = /(?<year>\d{4})-(?<month>\d{2})/.exec("due 2026-10");7 8  const pairs = [..."a1 b22 c333".matchAll(/([a-z])(\d+)/g)].map(9    (m) => `${m[1]}:${m[2]}`,10  );11 12  const userInput = "1+1";13  const escaped = userInput.replace(/[.*+?^${}()|[\]\\]/g, "\\$&");14 15  return {16    sharedGlobal,17    greedy: html.match(/<.+>/)[0],18    lazy: html.match(/<.+?>/)[0],19    namedGroups: { ...date.groups },20    matchWithG: "a1 b22".match(/\d+/g),21    pairs,22    swapped: "2026-10-03".replace(/(\d+)-(\d+)-(\d+)/, "$3/$2/$1"),23    doubled: "3 apples, 12 pears".replace(/\d+/g, (n) => String(n * 2)),24    unescaped: new RegExp(userInput).test("1+1"),25    escapedSource: escaped,26    escapedWorks: new RegExp(escaped).test("1+1"),27  };28}

Common mistakes

• str.length as a character count. It counts UTF-16 code units; use [...str].length for code points.
• A shared /.../g regex used with test. Results alternate between true and false for the same input, as the second example shows.
• replace("-", "+") expecting every dash replaced. A string pattern replaces the first occurrence only.
• Greedy .+ across delimiters. /<.+>/ on <b>x</b> matches the whole string; use .+? or a negated class [^>]+.

What interviewers ask

• Why does "👍".length return 2? — Strings are UTF-16; that emoji is one code point stored as a surrogate pair of two code units.
• What does the g flag change about test and exec? — They become stateful through , continuing from the previous match.