PHP Random String Generator: Under the Hood

How does a random string generator in PHP actually work? Looking under the hood

Have you ever wondered what happens when you click "Generate Password" on a website? Or when the system creates a unique identifier for your session? In nine out of ten cases, behind this is a PHP random string generator. It seems simple enough: take an array of characters, shuffle them, output the result. But the devil, as always, is in the details.

On Stack Overflow, the question "PHP random string generator" has thousands of views. And this is no accident. Beginners write bicycles with rand(), experienced developers use random_int(), and gurus argue about entropy. Let's figure out how it all really works and write our own ideal generator.

Why is rand() evil? (spoiler: predictability)

The most popular "beginner" method is to use rand() in a loop:

<?phpfunction generateRandomString($length = 10) {    $characters = '0123456789abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ';    $charactersLength = strlen($characters);    $randomString = '';    for ($i = 0; $i < $length; $i++) {        $randomString .= $characters[rand(0, $charactersLength - 1)];    }    return $randomString;}echo generateRandomString(); // Output: "aB3kL9xQ2m"?>

The code works. But only if you don't care about security. The rand() function uses a linear congruential generator (LCG). This is a deterministic algorithm: knowing the current state, you can predict all subsequent numbers. Moreover, rand() is initialized with a seed of 1 by default if you haven't called srand(). And if you have, the seed is often taken from time(), which gives only 2-3 million possibilities. Brute-forcing takes seconds.

Conclusion: rand() is suitable for games and random avatar selection, but is categorically unsuitable for generating tokens, passwords, or API keys.

Introducing mt_rand() — faster, but not more secure

PHP 4 introduced mt_rand(), based on the Mersenne Twister. It is faster and has a better period (2^19937 vs 2^31 for rand). Many still use it as an "advanced" version:

<?phpfunction generateTokenMtRand($length = 32) {    $chars = 'abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789';    $token = '';    for ($i = 0; $i < $length; $i++) {        $token .= $chars[mt_rand(0, strlen($chars) - 1)];    }    return $token;}?>

But there's a catch here too. The Mersenne Twister is cryptographically insecure. From a sequence of 624 generated numbers, you can reconstruct the internal state of the generator and predict all future values. This is more than enough for an attack on a password reset token.

The moral: mt_rand() is an excellent choice for simulations and tests, but not for security.

The Holy Trinity: random_int(), CSPRNG, and Entropy

Starting with PHP 7.0, we have random_int() and random_bytes(). These are cryptographically secure pseudo-random number generators (CSPRNG). What does this mean in practice?

• Entropy: PHP collects "noise" from the system: time between keystrokes, mouse movements, network packets, disk interrupts. Linux uses /dev/urandom, Windows uses CryptGenRandom.
• Unpredictability: Even knowing the algorithm, you cannot reconstruct the seed or predict the next number.
• Uniform distribution: Each byte occurs with a probability of exactly 1/256.

Here's what a correct PHP random string generator looks like