README

======

Full Documentation

A small PHP class to generate YouTube-like ids from numbers. Read documentation at http://hashids.org/php

Installation

You can install Hashids thru Composer (packagist has hashids/hashids package). In your composer.json file use:

{
    "require": {
        "alexanderzon/hashids": "2.0.*"
    }
}

And run: php composer.phar install. After that you can require the autoloader and use Hashids:

<?php

require_once 'vendor/autoload.php'
$hashids = new Hashids\Hashids('this is my salt');

Updating from v0.3 to 1.0?

Read the CHANGELOG at the bottom of this readme!

Example Usage

The simplest way to use Hashids:

<?php

$hashids = new Hashids\Hashids();

$id = $hashids->encode(1, 2, 3);
$numbers = $hashids->decode($id);

var_dump($id, $numbers);

string(5) "laHquq"
array(3) {
  [0]=>
  int(1)
  [1]=>
  int(2)
  [2]=>
  int(3)
}

And an example with all the custom parameters provided (unique salt value, minimum id length, custom alphabet):

<?php

$hashids = new Hashids\Hashids('this is my salt', 8, 'abcdefghij1234567890');

$id = $hashids->encode(1, 2, 3);
$numbers = $hashids->decode($hash);

var_dump($id, $numbers);

string(5) "514cdi42"
array(3) {
  [0]=>
  int(1)
  [1]=>
  int(2)
  [2]=>
  int(3)
}

Curses! #$%@

This code was written with the intent of placing created ids in visible places - like the URL. Which makes it unfortunate if generated hashes accidentally formed a bad word.

Therefore, the algorithm tries to avoid generating most common English curse words. This is done by never placing the following letters next to each other:

c, C, s, S, f, F, h, H, u, U, i, I, t, T

Big Numbers

Each number passed to the constructor cannot be negative or greater than 1 billion by default (1,000,000,000). Hashids encode() function will return an empty string if at least one of the numbers is out of bounds. Be sure to check for that -- no exception is thrown.

PHP starts approximating numbers when it does arithmetic on large integers (by converting them to floats). Which is usually not a big issue, but a problem when precise integers are needed.

However, if you have either GNU Multiple Precision --with-gmp, or BCMath Arbitrary Precision Mathematics --enable-bcmath libraries installed, Hashids will increase its upper limit to PHP_INT_MAX which is int(2147483647) on 32-bit systems and int(9223372036854775807) on 64-bit.

It will then use regular arithmetic on numbers less than 1 billion (because it's faster), and one of these libraries if greater than. GMP takes precedence over BCMath.

You can get the upper limit by doing: $hashids->get_max_int_value(); (which will stay at 1 billion if neither of the libraries is installed).

Speed

Even though speed is an important factor of every hashing algorithm, primary goal here was encoding several numbers at once while avoiding collisions.

On a 2.26 GHz Intel Core 2 Duo with 8GB of RAM, it takes about:

1. 0.000093 seconds to encode one number.
2. 0.000240 seconds to decode one id (while ensuring that it's valid).
3. 0.493436 seconds to generate 10,000 ids in a for loop.

On a 2.7 GHz Intel Core i7 with 16GB of RAM, it takes roughly:

1. 0.000067 seconds to encode one number.
2. 0.000113 seconds to decode one id (and ensuring that it's valid).
3. 0.297426 seconds to generate 10,000 ids in a for loop.

Sidenote: The numbers tested with were relatively small -- if you increase them, the speed will obviously decrease.

Notes

• If you want to squeeze out even more performance, set a shorter alphabet. Hashes will be less random and longer, but calculating them will be faster.

Changelog

2.0.0:

• The decode function will return an string or integer when you gives to the encode only one element (thanks @AlexanderZon)

1.0.5:

• bug fix for passing empty array to encode (thanks @bpahan)

1.0.3 & 1.0.4:

• adjusting examples (thanks @Trismegiste)
• proper version bump in const VERSION

1.0.2

• PSR-2 cleanup + interface changes (thanks @Trismegiste)
• encode() can accept array of integers (thanks @leunggamciu)

1.0.1

• bug fix for encode_hex() (thanks @leihog)
• unit test for encode_hex()/decode_hex()

1.0.0

• Several public functions are renamed to be more appropriate:

  • Function encrypt() changed to encode()
  • Function decrypt() changed to decode()
  • Function encrypt_hex() changed to encode_hex()
  • Function decrypt_hex() changed to decode_hex()

  Hashids was designed to encode integers, primary ids at most. We've had several requests to encrypt sensitive data with Hashids and this is the wrong algorithm for that. So to encourage more appropriate use, encrypt/decrypt is being "downgraded" to encode/decode.

• Version tag added: 1.0

• README.md updated

0.3.1

• Added encrypt_hex() and decrypt_hex() support
• Minor: Relaxed integer check in encrypt() function (can now pass strings of numbers)

0.3.0 - Warning: Hashes change in this version:

• Bug fix: better handling of big numbers: vinkla#3 (thanks @tobsn and @miquelfire)
• Bug fix: exception throwing in constructor
• Default maximum number is set to 1 billion: 1,000,000,000. Unless you have GNU Multiple Precision or BCMath Arbitrary Precision Mathematics library installed - then PHP_INT_MAX is used.
• Cleanup: private variables use underscores

0.2.1

• General directory cleanup + improvements
• Now only one library file for both PHP 5.3 and PHP 5.4
• Constants uppercased
• Namespace Hashids added to library class

0.2.0 - Warning: Hashes change in this version:

• Overall approximately 4x faster
• Consistent shuffle function uses slightly modified version of Fisher–Yates algorithm
• Generate large hash strings faster (more than 1000 chars)
• When using minimum hash length parameter, hash character disorder has been improved
• Basic English curse words will now be avoided even with custom alphabet
• Class name changed from hashids to Hashids
• New unit tests with PHPUnit (requires latest PHP)
• Composer package at packagist: https://packagist.org/packages/hashids/hashids
• Minor: a bit smaller code overall -- more motivation to port to other languages :P

0.1.3 - Warning: Hashes change in this version:

• Updated default alphabet (thanks to @speps)
• Constructor removes duplicate characters for default alphabet as well (thanks to @speps)

0.1.2 - Warning: Hashes change in this version:

• Minimum hash length can now be specified
• Added more randomness to hashes
• Added unit tests
• Added example files
• Changed warnings that can be thrown
• Renamed encode/decode to encrypt/decrypt
• Consistent shuffle does not depend on md5 anymore
• Speed improvements

0.1.1

• Speed improvements
• Bug fixes

0.1.0

• First commit

Contact

I am on the internets @IvanAkimov

License

MIT License. See the LICENSE file. You can use Hashids in open source projects and commercial products. Don't break the Internet. Kthxbye.