Random 64 character string generator

const ALPHABET = "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789";  // 62 characters

function randomString(length = 64) {
  // Largest multiple of 62 that fits in a uint32. Anything at or above it
  // is discarded, which is what keeps the mapping uniform.
  const limit = Math.floor(2 ** 32 / ALPHABET.length) * ALPHABET.length;
  const buf = new Uint32Array(1);
  let out = "";
  for (let i = 0; i < length; i++) {
    do { crypto.getRandomValues(buf); } while (buf[0] >= limit);
    out += ALPHABET[buf[0] % ALPHABET.length];
  }
  return out;
}

randomString();  // 64 characters

Without the rejection step, 2^32 does not divide evenly by 62 and the first four characters of the alphabet come up slightly more often.

import secrets
import string

ALPHABET = string.ascii_letters + string.digits  # 62 characters

"".join(secrets.choice(ALPHABET) for _ in range(64))

secrets.choice goes through randbelow, which rejects out-of-range values internally. random.choice is unbiased as well, but it runs on the Mersenne Twister and must not be used here.

import (
	"crypto/rand"
	"math/big"
)

const alphabet = "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789"

func RandomString(n int) (string, error) {
	b := make([]byte, n)
	max := big.NewInt(int64(len(alphabet)))
	for i := range b {
		v, err := rand.Int(rand.Reader, max)
		if err != nil {
			return "", err
		}
		b[i] = alphabet[v.Int64()]
	}
	return string(b), nil
}

RandomString(64)

crypto/rand.Int already does rejection sampling, so the result is uniform. math/rand is not a substitute at any size.

function random_string(int $length): string {
    $alphabet = 'ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789';
    $max = strlen($alphabet) - 1;
    $out = '';
    for ($i = 0; $i < $length; $i++) {
        $out .= $alphabet[random_int(0, $max)];
    }
    return $out;
}

random_string(64);

random_int is CSPRNG-backed and unbiased. rand(), mt_rand() and array_rand() are none of those things.

import java.security.SecureRandom;

static final String ALPHABET =
    "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789";
static final SecureRandom RNG = new SecureRandom();

StringBuilder sb = new StringBuilder(64);
for (int i = 0; i < 64; i++) {
    sb.append(ALPHABET.charAt(RNG.nextInt(ALPHABET.length())));
}
sb.toString();

nextInt(bound) rejects and redraws when the bound does not divide the range evenly, so no bias is introduced here.

using System.Security.Cryptography;

const string Alphabet =
    "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789";

var chars = new char[64];
for (int i = 0; i < chars.Length; i++)
{
    chars[i] = Alphabet[RandomNumberGenerator.GetInt32(Alphabet.Length)];
}
string value = new string(chars);

RandomNumberGenerator.GetInt32 is unbiased. System.Random is not cryptographic and is the wrong class for this job.

# Linux and macOS. LC_ALL=C stops tr rejecting bytes outside the locale.
LC_ALL=C tr -dc 'A-Za-z0-9' < /dev/urandom | head -c 64; echo

# openssl, if you want 64 hexadecimal characters instead
# (32 bytes = 64 hex characters = 256 bits, exactly)
openssl rand -hex 32

tr -dc discards bytes outside the set rather than folding them into it, so the one-liner is uniform. /dev/urandom is the kernel CSPRNG on both platforms.