README

Pack and unpack binary integers and floats in any byte order.

Installation

composer require thesis/endian

Usage

use Thesis\Endian;

// Big endian (= network byte order)
$bytes = Endian\Order::Big->packInt32(-200);
$value = Endian\Order::Big->unpackInt32($bytes); // -200
$bytes = Endian\Order::Network->packInt32(-200);
$value = Endian\Order::Network->unpackInt32($bytes); // -200

// Little endian
$bytes = Endian\Order::Little->packFloat(2.2);
$value = Endian\Order::Little->unpackFloat($bytes); // 2.2

// Native byte order of the current machine
$order = Endian\Order::native(); // Order::Big or Order::Little

Design decisions

No narrow int types on input

Pack methods accept plain int rather than narrow PHPStan types like Int8 or Int32. This is intentional: requiring callers to carry and assert narrow types would flood driver code with redundant checks and type imports. Thus, we've decided to keep the validation on our side.

Bounds are checked with assert(), which means zero overhead in production when assertions are disabled (zend.assertions = -1).

No object wrappers for 8/16/32-bit

8/16/32-bit integers are represented as native PHP int, not value objects. This avoids allocation and method-call overhead on every pack/unpack — important in tight loops typical of binary protocol parsing.

64-bit integers

64-bit values use BcMath\Number to handle the full unsigned range beyond PHP_INT_MAX:

use Thesis\Endian;
use BcMath\Number;

$bytes = Endian\Order::Big->packUint64(new Number('18446744073709551615'));
$value = Endian\Order::Big->unpackUint64($bytes); // 18446744073709551615

Supported types