Emoji Encoding Guide: UTF-8, UTF-16 & Surrogate Pairs

Why Emoji Break Your String.length

If you've ever been surprised that '😀'.length === 2 in JavaScript, you've encountered one of the most common emoji encoding pitfalls. This guide explains why it happens and how to handle emoji correctly in code.

The Three Unicode Encodings

Unicode defines three encoding forms. Each uses a different strategy to convert code points into bytes:

UTF-8UTF-8
Кодировка Unicode переменной ширины, использующая от 1 до 4 байт на символ; является доминирующей в интернете (применяется более чем на 98% сайтов).: The Web Standard

UTF-8 uses 1-4 bytes per character and is backward-compatible with ASCII:

Every emoji requires 4 bytes in UTF-8 because emoji code points are above U+FFFF. A ZWJОбъединитель нулевой ширины (ZWJ)
Невидимый символ Unicode (U+200D), используемый для соединения нескольких эмодзи в одно составное, например при создании эмодзи профессий из фигур людей и предметов. sequence like 👩‍💻 uses 11 bytes (4 + 3 + 4 for person + ZWJ + laptop).

UTF-16UTF-16
Кодировка Unicode переменной ширины, использующая 2 или 4 байта на символ; применяется внутри JavaScript, Java и Windows.: JavaScript and Java's Native Encoding

UTF-16 uses 2 or 4 bytes per character:

Characters above U+FFFF — including virtually all emoji — require a surrogate pair: two 16-bit code units that encode one code point.

UTF-32UTF-32
Кодировка Unicode фиксированной ширины, использующая ровно 4 байта на символ; обеспечивает прямое отображение кодовых точек за счёт увеличенного размера файла.: Simple but Wasteful

UTF-32 uses exactly 4 bytes per code point. Simple for processing (string[i] always gives you one code point), but uses 4x the memory of ASCII text.

Surrogate Pairs Explained

Surrogate pairs are the key to understanding JavaScript emoji behavior. Here's the math:

Code Point: U+1F600 (😀)
Offset: 0x1F600 - 0x10000 = 0xF600

High Surrogate: 0xD800 + (0xF600 >> 10) = 0xD800 + 0x3D = 0xD83D
Low Surrogate:  0xDC00 + (0xF600 & 0x3FF) = 0xDC00 + 0x200 = 0xDE00

Result: 0xD83D 0xDE00

This is why '😀'.charCodeAt(0) returns 55357 (0xD83D) — the high surrogate — and '😀'.charCodeAt(1) returns 56832 (0xDE00) — the low surrogate.

Common Pitfalls

1. String Length

// WRONG: counts UTF-16 code units
'😀'.length              // 2
'👨‍👩‍👧'.length           // 8

// CORRECT: counts grapheme clusters
[...new Intl.Segmenter().segment('👨‍👩‍👧')].length  // 1

2. String Slicing

// WRONG: splits surrogate pair
'Hello 😀'.slice(0, 7)  // 'Hello \uD83D' (broken!)

// CORRECT: use spread or Array.from
[...'Hello 😀'].slice(0, 7).join('')  // 'Hello 😀'

3. Regular Expressions

// WRONG: . doesn't match emoji by default
/^.$/.test('😀')  // false

// CORRECT: use u flag for Unicode awareness
/^.$/u.test('😀')  // true

4. Database Storage

When using MySQL, ensure your column uses utf8mb4 (not utf8 which only supports 3-byte characters). PostgreSQL handles this correctly by default with its TEXT type.

-- MySQL: must use utf8mb4 for emoji
ALTER TABLE posts MODIFY content TEXT CHARACTER SET utf8mb4;

Language-Specific Tips

Python 3

Python 3 handles emoji gracefully — len('😀') returns 1 because Python uses code points internally.

emoji = '👩‍💻'
len(emoji)           # 3 (code points: woman + ZWJ + laptop)
emoji.encode('utf-8') # b'\xf0\x9f\x91\xa9\xe2\x80\x8d\xf0\x9f\x92\xbb'

Java

Java strings are UTF-16, like JavaScript:

"😀".length()          // 2 (surrogate pair)
"😀".codePointCount(0, "😀".length())  // 1 (actual code point count)

Rust

Rust strings are UTF-8 by default and distinguish between bytes, chars, and graphemes:

"😀".len()            // 4 (bytes)
"😀".chars().count()  // 1 (code points)

Analyze Any Emoji's Encoding

Use our Sequence Analyzer to see the complete encoding breakdown of any emoji — UTF-8 bytes, UTF-16 surrogates, code points, and component roles.