PHP Internationalization & Encoding

UTF-8 Everywhere

// Set early — affects all mb_* functions mb_internal_encoding('UTF-8'); header('Content-Type: text/html; charset=UTF-8');

// Use mb_* for all string operations on user text $len = mb_strlen($str); // NOT strlen() — counts bytes, not chars $upper = mb_strtoupper($str); // NOT strtoupper() $sub = mb_substr($str, 0, 10); // NOT substr() $pos = mb_strpos($str, $needle);

// PHP 8.4+ multibyte trim $clean = mb_trim($str); $clean = mb_ltrim($str); $clean = mb_rtrim($str);

Rule Detail

strlen() counts bytes "\xF0\x9F\x9A\x80" = 4 bytes, 1 character — use mb_strlen()

substr() corrupts multibyte Use mb_substr() for slicing

strtoupper() / strtolower()

Only handles ASCII — use mb_strtoupper()

Set default_charset = UTF-8

In php.ini — replaces deprecated mbstring.internal_encoding

MySQL needs utf8mb4

utf8 is only 3 bytes, can't store emoji

Encoding Detection & Conversion

// Convert from unknown encoding to UTF-8 $utf8 = mb_convert_encoding($str, 'UTF-8', 'ISO-8859-1');

// Detect encoding (unreliable — prefers explicit metadata) $enc = mb_detect_encoding($str, ['UTF-8', 'ISO-8859-1', 'Windows-1252'], true);

// Validate UTF-8 (strict mode returns false for invalid) if (mb_detect_encoding($str, 'UTF-8', true) === false) { $str = mb_convert_encoding($str, 'UTF-8', 'Windows-1252'); }

Gotcha Detail

mb_detect_encoding() guesses Same bytes can be valid in multiple encodings

Always declare encoding HTTP headers, HTML meta, DB connection charset

iconv() varies by system GNU libiconv is more reliable than glibc

Regex needs /u modifier Without it, PCRE treats input as single-byte

Database & JSON Encoding

// MySQL: always use utf8mb4 $pdo = new PDO('mysql:host=localhost;dbname=app;charset=utf8mb4', $user, $pass);

// PostgreSQL $pdo = new PDO("pgsql:host=localhost;dbname=app;options='-c client_encoding=UTF8'");

// JSON: all strings must be UTF-8 $json = json_encode($data, JSON_UNESCAPED_UNICODE | JSON_THROW_ON_ERROR);

// Handle broken UTF-8 in JSON (PHP 7.2+) $json = json_encode($data, JSON_INVALID_UTF8_SUBSTITUTE); // replaces with U+FFFD

intl Extension (ICU)

// Locale-aware string comparison $coll = Collator::create('de_DE'); $coll->compare('a with umlaut', 'z'); // locale-correct ordering $coll->sort($array);

// Number formatting $fmt = NumberFormatter::create('de_DE', NumberFormatter::DECIMAL); echo $fmt->format(1234.56); // "1.234,56" $fmt->formatCurrency(99.99, 'EUR'); // "99,99 EUR"

// Date formatting $fmt = IntlDateFormatter::create('fr_FR', IntlDateFormatter::LONG, IntlDateFormatter::SHORT); echo $fmt->format(time()); // "1 fevrier 2026 a 14:30"

// Pluralization $msg = '{count, plural, =0{no items} one{# item} other{# items}}'; echo MessageFormatter::formatMessage('en_US', $msg, ['count' => 5]); // "5 items"

// Transliteration $t = Transliterator::create('Any-Latin; Latin-ASCII'); echo $t->transliterate('Privet mir'); // transliterated output

Class Use for

Collator

Locale-aware sorting and comparison

NumberFormatter

Numbers, currencies, percentages

IntlDateFormatter

Dates/times per locale

MessageFormatter

ICU messages with plurals/gender

Transliterator

Script conversion (Cyrillic to Latin, etc.)

Normalizer

Unicode normalization (NFC, NFD, NFKC, NFKD)

HTTP Charset Pipeline

Input -> Convert to UTF-8 -> Process with mb_* -> Output with charset header -> Store as utf8mb4

Stage Action

Input Detect/convert: mb_convert_encoding($input, 'UTF-8', $detected)

Processing Use mb_* functions; regex with /u modifier

HTML output header('Content-Type: text/html; charset=UTF-8')

JSON output json_encode($data, JSON_UNESCAPED_UNICODE)

Database charset=utf8mb4 in DSN; SET NAMES utf8mb4

Files Verify encoding on read; BOM handling if needed