uniqid

(PHP 4, PHP 5, PHP 7, PHP 8)

uniqid生成一个唯一ID

说明

uniqid ( string $prefix = "" , bool $more_entropy = false ) : string

获取一个带前缀、基于当前时间微秒数的唯一ID。

Caution

本函数并不会生成安全加密的值,不应用于加密用途。若需要安全加密的值,考虑使用 random_int()random_bytes()openssl_random_pseudo_bytes() 替代。

Warning

此函数不保证返回值的唯一性。 由于绝大多数系统使用 NTP 或者类似服务调整系统的时间,所以系统时间经常发生变化。 此外,进程/线程可能不会返回唯一的 ID。 用 more_entropy 来增加唯一性的概率。

参数

prefix

有用的参数。例如:如果在多台主机上可能在同一微秒生成唯一ID。

prefix为空,则返回的字符串长度为13。more_entropytrue,则返回的字符串长度为23。

more_entropy

如果设置为 trueuniqid() 会在返回的字符串结尾增加额外的熵(使用combined linear congruential generator)。 使得唯一ID更具唯一性。

返回值

返回字符串形式的唯一ID。

Warning

此函数努力创建唯一识别符,但它不保证返回值得唯一性。

范例

Example #1 uniqid() 例子

<?php
/* A uniqid, like: 4b3403665fea6 */
printf("uniqid(): %s\r\n"uniqid());

/* We can also prefix the uniqid, this the same as 
 * doing:
 *
 * $uniqid = $prefix . uniqid();
 * $uniqid = uniqid($prefix);
 */
printf("uniqid('php_'): %s\r\n"uniqid('php_'));

/* We can also activate the more_entropy parameter, which is 
 * required on some systems, like Cygwin. This makes uniqid()
 * produce a value like: 4b340550242239.64159797
 */
printf("uniqid('', true): %s\r\n"uniqid(''true));
?>

注释

Note:

在Cygwin环境下,为了使此函数能够工作,more_entropy 必须设置为 true

User Contributed Notes

mt at banita dot de 18-Jan-2021 11:22
note hackan at gmail dot com is not or no longer valid even in a for loop uniqid delivers uniq values

Tested:
PHP 7.3.18 (cli) (built: May 12 2020 10:55:29) ( ZTS MSVC15 (Visual C++ 2017) x64 )
PHP 7.4.3 (cli) (built: Oct  6 2020 15:47:56) ( NTS )
DimeCadmium 25-Jul-2019 07:00
Most of these notes are grossly wrong.

First: you probably shouldn't be using uniqid at all. I can't think of a good reason to use it. Maybe to use as an identifier for a particular run of your script in logging, where you don't particularly care if there's a collision? I don't know.

Second: don't rely on the output of uniqid to be, in any way, unique. This will break if you use it twice in quick succession, or if you have two different users at the same time, or if the phase of the moon is wrong.

Third: to nodkz who wrote the clientIPToHex function - you reinvented so many wheels. bin2hex(inet_pton($ip)) will do the same thing.

Fourth: do not use random bytes for a unique identifier (like hackan suggests). That only serves to virtually guarantee that you will eventually have a collision. (You can use random bytes, and get another set of random bytes if there's a collision... but it's a lot easier to just use a sequential identifier like you should, possibly along with a random key)
hackan at gmail dot com 05-Nov-2016 07:17
Seriously, avoid using this function. Here's an example of why:

<?php
for($i=0;$i<20;$i++) {
    echo
uniqid(), PHP_EOL;
}
?>

Output:

5819f3ad1c0ce
5819f3ad1c0ce
5819f3ad1c0ce
5819f3ad1c0ce
5819f3ad1c0ce
5819f3ad1c0ce
5819f3ad1c0ce
5819f3ad1c0ce
5819f3ad1c4b6
5819f3ad1c4b6
5819f3ad1c4b6
5819f3ad1c4b6
5819f3ad1c4b6
5819f3ad1c4b6
5819f3ad1c4b6
5819f3ad1c4b6
5819f3ad1c4b6
5819f3ad1c4b6
5819f3ad1c4b6
5819f3ad1c4b6

As you can see, using it w/ a DB can cause the creation of documents with repeated ID's. You should instead opt for:

<?php
function uniqidReal($lenght = 13) {
   
// uniqid gives 13 chars, but you could adjust it to your needs.
   
if (function_exists("random_bytes")) {
       
$bytes = random_bytes(ceil($lenght / 2));
    } elseif (
function_exists("openssl_random_pseudo_bytes")) {
       
$bytes = openssl_random_pseudo_bytes(ceil($lenght / 2));
    } else {
        throw new
Exception("no cryptographically secure random function available");
    }
    return
substr(bin2hex($bytes), 0, $lenght);
}
?>

Test:
<?php
for($i=0;$i<20;$i++) {
    echo
uniqid(), "\t", uniqidReal(), PHP_EOL;
}
?>

Output:

5819fa0b63be3   9f39aa0ecd89d
5819fa0b70ed3   2c0735cabfcce
5819fa0b712bb   15e45d1ca1e90
5819fa0b712bb   89593dc230eb3
5819fa0b712bb   449795704aeef
5819fa0b712bb   b046877b80ac9
5819fa0b712bb   0a6fa0ae3ec7b
5819fa0b712bb   ba2f3f4d6afe0
5819fa0b712bb   af03cfac83fd6
5819fa0b712bb   eb9c3c6d475c0
5819fa0b712bb   edfbbf59d5e1b
5819fa0b712bb   500dca18888d4
5819fa0b716a3   4f5a40ef715f1
5819fa0b716a3   154e42b616825
5819fa0b716a3   a879a22663c9b
5819fa0b716a3   ea7c044ddda8a
5819fa0b716a3   2c81a44dc674e
5819fa0b716a3   bb32f37304fd9
5819fa0b716a3   30cdf6c0317d7
5819fa0b716a3   d25f529d126ae
php at dot dot dot metehanarslan dot com 21-Apr-2016 12:33
Here is my approach to generate short string unique id. I am just increasing base of UUID. If you need even shorter keys you may add more chars to $index . When $index gets longer $out will be shorter.

<?php
function struuid($entropy)
{
   
$s=uniqid("",$entropy);
   
$num= hexdec(str_replace(".","",(string)$s));
   
$index = '1234567890ABCDEFGHIJKLMNOPQRSTUVWXYZ';
   
$base= strlen($index);
   
$out = '';
        for(
$t = floor(log10($num) / log10($base)); $t >= 0; $t--) {
           
$a = floor($num / pow($base,$t));
           
$out = $out.substr($index,$a,1);
           
$num = $num-($a*pow($base,$t));
        }
    return
$out;
}

echo
struuid(false); //Return sample: F4518NTQTQ
echo struuid(true);  //Return sample: F451FAHSUCD90N6YNRBQHLZ9E1W

?>
jensenv at cshs dot org 01-May-2015 12:26
DUPLICATE UID WARNING: Be aware that calling this function many times in a tight loop (for instance, to assign UIDs to objects in an array) can result in many of the UIDs being identical, since less than a microsecond may have passed since the previous call. One simple solution is to compare the latest value returned with the previous result, and keep calling in a loop until the new value is different, if you don't mind this call causing a delay of 1 microsecond per call when necessary.
ciantic at NOSPAM dot oksidi dot com 05-Dec-2013 03:34
If you are storing the value to database it might be more efficient to use more letters than in hexadecimal, for me I decided 0-9a-z is good enough.

Notice also that many databases are case insensitive by default so using capitals maybe unwise anyway:

<?php

function uniqid_base36($more_entropy=false) {
   
$s = uniqid('', $more_entropy);
    if (!
$more_entropy)
        return
base_convert($s, 16, 36);
       
   
$hex = substr($s, 0, 13);
   
$dec = $s[13] . substr($s, 15); // skip the dot
   
return base_convert($hex, 16, 36) . base_convert($dec, 10, 36);
}

echo
uniqid_base36() . "\n"; // ?eb98xzzhq7
echo uniqid_base36(true) . "\n"; // eb98xzzhr8dervre

?>

String length may vary with this method.
Anonymous 22-Nov-2012 06:52
I wouldn't use:
$prefix = exec("hostname");
- since it's CPU-expensive.

Instead use php_uname('n') if you have many visitors.
redtraider at gmail dot com 14-Feb-2012 12:37
I use this function to generate microsoft-compatible GUID's.

<?php
 
public function create_guid($namespace = '') {    
    static
$guid = '';
   
$uid = uniqid("", true);
   
$data = $namespace;
   
$data .= $_SERVER['REQUEST_TIME'];
   
$data .= $_SERVER['HTTP_USER_AGENT'];
   
$data .= $_SERVER['LOCAL_ADDR'];
   
$data .= $_SERVER['LOCAL_PORT'];
   
$data .= $_SERVER['REMOTE_ADDR'];
   
$data .= $_SERVER['REMOTE_PORT'];
   
$hash = strtoupper(hash('ripemd128', $uid . $guid . md5($data)));
   
$guid = '{' .  
           
substr($hash08) .
           
'-' .
           
substr($hash84) .
           
'-' .
           
substr($hash, 124) .
           
'-' .
           
substr($hash, 164) .
           
'-' .
           
substr($hash, 20, 12) .
           
'}';
    return
$guid;
  }
?>
tom 17-Jun-2011 12:40
mt_rand() features in a lot of comments here. It is a good PRNG for Monte Carlo simulations, not for anything related to security. Wikipedia's page on cryptographically-secure PRNGs explains. If you want it to be hard for an attacker to guess or predict a "random" UUID, try using /dev/random instead.

<?php
$r
= unpack('v*', fread(fopen('/dev/random', 'r'),16));
$uuid = sprintf('%04x%04x-%04x-%04x-%04x-%04x%04x%04x',
   
$r[1], $r[2], $r[3], $r[4] & 0x0fff | 0x4000,
   
$r[5] & 0x3fff | 0x8000, $r[6], $r[7], $r[8]);
?>

That's obviously not production code. It's just for illustration.
Anonymous 16-Jun-2011 04:08
Prefix can be useful, for instance, if you generate identifiers simultaneously on several hosts that might happen to generate the identifier at the same microsecond.
So we can include the hostname / servername in the id.
<?php
echo uniqid(php_uname('n'), true);
// Output: darkstar4dfa8c27aea106.40781203
?>
rommel at rommelsantor dot com 28-May-2011 03:29
The php5-uuid functions could definitely use some documentation to clarify how they should be used, but here's what I've gleaned by examining the OSSP source code (found here: http://ossp-uuid.sourcearchive.com/documentation/1.5.1-1ubuntu1/php_2uuid_8c-source.html).

The uuid_make() function takes two arguments when generating v1 or v4, but four arguments are required when generating v3 or v5. The first two arguments have been demonstrated below and are straightforward, so I'll skip to the as-yet non-described arguments.

The third argument to uuid_make() is: $namespace
  - this is a secondary resource created with uuid_create(); it is apparently used to generate an internal UUID, which is used as the namespace of the output UUID

The fourth argument to uuid_make() is: $url
  - this is the value that is to be hashed (MD5 for v3, SHA-1 for v5); it may be any string or even null

Here's a simple class illustrating the proper usage (note that if php5-uuid is not installed on your system, each function call will just return false):

<?php
class UUID {
 
/**
   * Generates version 1: MAC address
   */
 
public static function v1() {
    if (!
function_exists('uuid_create'))
      return
false;

   
uuid_create(&$context);
   
uuid_make($context, UUID_MAKE_V1);
   
uuid_export($context, UUID_FMT_STR, &$uuid);
    return
trim($uuid);
  }

 
/**
   * Generates version 3 UUID: MD5 hash of URL
   */
 
public static function v3($i_url) {
    if (!
function_exists('uuid_create'))
      return
false;

    if (!
strlen($i_url))
     
$i_url = self::v1();

   
uuid_create(&$context);
   
uuid_create(&$namespace);

   
uuid_make($context, UUID_MAKE_V3, $namespace, $i_url);
   
uuid_export($context, UUID_FMT_STR, &$uuid);
    return
trim($uuid);
  }

 
/**
   * Generates version 4 UUID: random
   */
 
public static function v4() {
    if (!
function_exists('uuid_create'))
      return
false;

   
uuid_create(&$context);

   
uuid_make($context, UUID_MAKE_V4);
   
uuid_export($context, UUID_FMT_STR, &$uuid);
    return
trim($uuid);
  }

 
/**
   * Generates version 5 UUID: SHA-1 hash of URL
   */
 
public static function v5($i_url) {
    if (!
function_exists('uuid_create'))
      return
false;

    if (!
strlen($i_url))
     
$i_url = self::v1();

   
uuid_create(&$context);
   
uuid_create(&$namespace);

   
uuid_make($context, UUID_MAKE_V5, $namespace, $i_url);
   
uuid_export($context, UUID_FMT_STR, &$uuid);
    return
trim($uuid);
  }
}
?>

And here's a demonstration:

<?php
for ($i = 1; $i <= 3; ++$i) {
  echo
'microtime = ' . microtime(true) . '<br/>';
  echo
"V1 UUID: " . UUID::v1() . '<br/>';
  echo
"V3 UUID of URL='abc': " . UUID::v3('abc') . '<br/>';
  echo
"V4 UUID: " . UUID::v4() . '<br/>';
  echo
"V5 UUID of URL=null: " . UUID::v5(null) . '<br/>';
  echo
'<hr/>';
}
?>

And the output:

microtime = 1306620716.0457
V1 UUID: 7fddae8e-8977-11e0-bc11-003048c3b1f2
V3 UUID of URL='abc': 522ec739-ca63-3ec5-b082-08ce08ad65e2
V4 UUID: b3851ec7-4871-4527-92b5-ef5616bae1e6
V5 UUID of URL=null: e129f27c-5103-5c5c-844b-cdf0a15e160d
-------------------
microtime = 1306620716.0465
V1 UUID: 7fddb83e-8977-11e0-9e6e-003048c3b1f2
V3 UUID of URL='abc': 522ec739-ca63-3ec5-b082-08ce08ad65e2
V4 UUID: 7e78fe0d-59b8-4637-af7f-e88d221a7d1e
V5 UUID of URL=null: e129f27c-5103-5c5c-844b-cdf0a15e160d
-------------------
microtime = 1306620716.0467
V1 UUID: 7fddbfb4-8977-11e0-a2bc-003048c3b1f2
V3 UUID of URL='abc': 522ec739-ca63-3ec5-b082-08ce08ad65e2
V4 UUID: 12a940c7-0f3f-46a1-bb5f-bdd602e10654
V5 UUID of URL=null: e129f27c-5103-5c5c-844b-cdf0a15e160d

As you can see, the calls to v3() always return the same UUID because the same URL parameter, "abc", is always supplied. The same goes for the v5() function which is always supplied a null URL.

The v4() UUIDs are always entirely different because they are (pseudo)random. And the v1() calls are very similar but just slightly different because it's based on the computer's MAC address and the current time.
php at ryanmckeel dot com 29-Jun-2010 10:15
Calls to uuid_make that use the constants UUID_MAKE_V5 or UUID_MAKE_V3 (using Debian package php5-uuid available June 2010) will not work with only two variables.

I could not find good documentation, so I read some source code and figured out that this would work:

uuid_create(&$v5);
//uuid_make($v5, UUID_MAKE_V5);
uuid_make($v5, UUID_MAKE_V5,$v5,$uniqid());
uuid_export($v5, UUID_FMT_STR, &$v5String);

Please use at your own risk.  This may not be the best way to give this param variables, but it at least makes it work what appears to be properly (generating unique ID's).

I imagine that UUID_MAKE_V3 is similar in what it needs.

  Ryan
Enrico Pallazzo 21-Mar-2010 02:45
Running:

rand_uniqid(9007199254740989);

will return 'PpQXn7COf' and:

rand_uniqid('PpQXn7COf', true);

will return '9007199254740989'

---

If you want the rand_uniqid to be at least 6 letter long, use the $pad_up = 6 argument

---

You can support even more characters (making the resulting rand_uniqid even smaller) by adding characters to the $index var at the top of the function body.

---

<?php
function rand_uniqid($in, $to_num = false, $pad_up = false, $passKey = null)
{
   
$index = "abcdefghijklmnopqrstuvwxyz0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZ";
    if (
$passKey !== null) {
       
// Although this function's purpose is to just make the
        // ID short - and not so much secure,
        // you can optionally supply a password to make it harder
        // to calculate the corresponding numeric ID

       
for ($n = 0; $n<strlen($index); $n++) {
           
$i[] = substr( $index,$n ,1);
        }

       
$passhash = hash('sha256',$passKey);
       
$passhash = (strlen($passhash) < strlen($index))
            ?
hash('sha512',$passKey)
            :
$passhash;

        for (
$n=0; $n < strlen($index); $n++) {
           
$p[] =  substr($passhash, $n ,1);
        }

       
array_multisort($pSORT_DESC, $i);
       
$index = implode($i);
    }

   
$base  = strlen($index);

    if (
$to_num) {
       
// Digital number  <<--  alphabet letter code
       
$in  = strrev($in);
       
$out = 0;
       
$len = strlen($in) - 1;
        for (
$t = 0; $t <= $len; $t++) {
           
$bcpow = bcpow($base, $len - $t);
           
$out   = $out + strpos($index, substr($in, $t, 1)) * $bcpow;
        }

        if (
is_numeric($pad_up)) {
           
$pad_up--;
            if (
$pad_up > 0) {
               
$out -= pow($base, $pad_up);
            }
        }
       
$out = sprintf('%F', $out);
       
$out = substr($out, 0, strpos($out, '.'));
    } else {
       
// Digital number  -->>  alphabet letter code
       
if (is_numeric($pad_up)) {
           
$pad_up--;
            if (
$pad_up > 0) {
               
$in += pow($base, $pad_up);
            }
        }

       
$out = "";
        for (
$t = floor(log($in, $base)); $t >= 0; $t--) {
           
$bcp = bcpow($base, $t);
           
$a   = floor($in / $bcp) % $base;
           
$out = $out . substr($index, $a, 1);
           
$in  = $in - ($a * $bcp);
        }
       
$out = strrev($out); // reverse
   
}

    return
$out;
}

echo
rand_uniqid(1);
?>
keith at keithtyler dot com 07-Dec-2009 06:18
For the record, the underlying function to uniqid() appears to be roughly as follows:

$m=microtime(true);
sprintf("%8x%05x\n",floor($m),($m-floor($m))*1000000);

In other words, first 8 hex chars = Unixtime, last 5 hex chars = microseconds. This is why it has microsecond precision. Also, it provides a means by which to reverse-engineer the time when a uniqid was generated:

date("r",hexdec(substr(uniqid(),0,8)));

Increasingly as you go further down the string, the number becomes "more unique" over time, with the exception of digit 9, where numeral prevalence is 0..3>4>5..f, because of the difference between 10^6 and 16^5 (this is presumably true for the remaining digits as well but much less noticeable).
Andrew Moore 04-Dec-2009 08:45
The following class generates VALID RFC 4211 COMPLIANT Universally Unique IDentifiers (UUID) version 3, 4 and 5.

Version 3 and 5 UUIDs are named based. They require a namespace (another valid UUID) and a value (the name). Given the same namespace and name, the output is always the same.

Version 4 UUIDs are pseudo-random.

UUIDs generated below validates using OSSP UUID Tool, and output for named-based UUIDs are exactly the same. This is a pure PHP implementation.

<?php
class UUID {
  public static function
v3($namespace, $name) {
    if(!
self::is_valid($namespace)) return false;

   
// Get hexadecimal components of namespace
   
$nhex = str_replace(array('-','{','}'), '', $namespace);

   
// Binary Value
   
$nstr = '';

   
// Convert Namespace UUID to bits
   
for($i = 0; $i < strlen($nhex); $i+=2) {
     
$nstr .= chr(hexdec($nhex[$i].$nhex[$i+1]));
    }

   
// Calculate hash value
   
$hash = md5($nstr . $name);

    return
sprintf('%08s-%04s-%04x-%04x-%12s',

     
// 32 bits for "time_low"
     
substr($hash, 0, 8),

     
// 16 bits for "time_mid"
     
substr($hash, 8, 4),

     
// 16 bits for "time_hi_and_version",
      // four most significant bits holds version number 3
     
(hexdec(substr($hash, 12, 4)) & 0x0fff) | 0x3000,

     
// 16 bits, 8 bits for "clk_seq_hi_res",
      // 8 bits for "clk_seq_low",
      // two most significant bits holds zero and one for variant DCE1.1
     
(hexdec(substr($hash, 16, 4)) & 0x3fff) | 0x8000,

     
// 48 bits for "node"
     
substr($hash, 20, 12)
    );
  }

  public static function
v4() {
    return
sprintf('%04x%04x-%04x-%04x-%04x-%04x%04x%04x',

     
// 32 bits for "time_low"
     
mt_rand(0, 0xffff), mt_rand(0, 0xffff),

     
// 16 bits for "time_mid"
     
mt_rand(0, 0xffff),

     
// 16 bits for "time_hi_and_version",
      // four most significant bits holds version number 4
     
mt_rand(0, 0x0fff) | 0x4000,

     
// 16 bits, 8 bits for "clk_seq_hi_res",
      // 8 bits for "clk_seq_low",
      // two most significant bits holds zero and one for variant DCE1.1
     
mt_rand(0, 0x3fff) | 0x8000,

     
// 48 bits for "node"
     
mt_rand(0, 0xffff), mt_rand(0, 0xffff), mt_rand(0, 0xffff)
    );
  }

  public static function
v5($namespace, $name) {
    if(!
self::is_valid($namespace)) return false;

   
// Get hexadecimal components of namespace
   
$nhex = str_replace(array('-','{','}'), '', $namespace);

   
// Binary Value
   
$nstr = '';

   
// Convert Namespace UUID to bits
   
for($i = 0; $i < strlen($nhex); $i+=2) {
     
$nstr .= chr(hexdec($nhex[$i].$nhex[$i+1]));
    }

   
// Calculate hash value
   
$hash = sha1($nstr . $name);

    return
sprintf('%08s-%04s-%04x-%04x-%12s',

     
// 32 bits for "time_low"
     
substr($hash, 0, 8),

     
// 16 bits for "time_mid"
     
substr($hash, 8, 4),

     
// 16 bits for "time_hi_and_version",
      // four most significant bits holds version number 5
     
(hexdec(substr($hash, 12, 4)) & 0x0fff) | 0x5000,

     
// 16 bits, 8 bits for "clk_seq_hi_res",
      // 8 bits for "clk_seq_low",
      // two most significant bits holds zero and one for variant DCE1.1
     
(hexdec(substr($hash, 16, 4)) & 0x3fff) | 0x8000,

     
// 48 bits for "node"
     
substr($hash, 20, 12)
    );
  }

  public static function
is_valid($uuid) {
    return
preg_match('/^\{?[0-9a-f]{8}\-?[0-9a-f]{4}\-?[0-9a-f]{4}\-?'.
                     
'[0-9a-f]{4}\-?[0-9a-f]{12}\}?$/i', $uuid) === 1;
  }
}

// Usage
// Named-based UUID.

$v3uuid = UUID::v3('1546058f-5a25-4334-85ae-e68f2a44bbaf', 'SomeRandomString');
$v5uuid = UUID::v5('1546058f-5a25-4334-85ae-e68f2a44bbaf', 'SomeRandomString');

// Pseudo-random UUID

$v4uuid = UUID::v4();
?>
John Haugeland from FullOf.BS 26-May-2009 10:11
Generating an MD5 from a unique ID is naive and reduces much of the value of unique IDs, as well as providing significant (attackable) stricture on the MD5 domain.  That's a deeply broken thing to do.  The correct approach is to use the unique ID on its own; it's already geared for non-collision.

IDs should never be obfuscated for security, so if you're worried about someone guessing your ID, fix the system, don't just make it harder to guess (because it's nowhere near as difficult to guess as you imagine: you can just brute force the 60,000 MD5s that are generatable from millisecond IDs over the course of a given minute, which the typical computer can do in less than 0.1s).

If you absolutely need to involve a hash somehow - maybe to placate a boss who thinks they understand security much better than they actually do - append it instead.

function BadIdeaID() { return uniqid() . '_' . md5(mt_rand()); }
sean at seancolombo dot com 06-Jan-2009 07:56
This builds slightly on david's post below.  The differences are that it doesn't require Cake anymore and there is a graceful fallback for /dev/urandom in case that isn't available (/dev/urandom is not available on windows systems for example).

Since it uses mt_rand(), it is still cryptographically secure.

<?php
/**
     * @brief Generates a Universally Unique IDentifier, version 4.
     *
     * This function generates a truly random UUID. The built in CakePHP String::uuid() function
     * is not cryptographically secure. You should uses this function instead.
     *
     * @see http://tools.ietf.org/html/rfc4122#section-4.4
     * @see http://en.wikipedia.org/wiki/UUID
     * @return string A UUID, made up of 32 hex digits and 4 hyphens.
     */
     
function uuidSecure() {
      
       
$pr_bits = null;
       
$fp = @fopen('/dev/urandom','rb');
        if (
$fp !== false) {
           
$pr_bits .= @fread($fp, 16);
            @
fclose($fp);
        } else {
           
// If /dev/urandom isn't available (eg: in non-unix systems), use mt_rand().
           
$pr_bits = "";
            for(
$cnt=0; $cnt < 16; $cnt++){
               
$pr_bits .= chr(mt_rand(0, 255));
            }
        }
      
       
$time_low = bin2hex(substr($pr_bits,0, 4));
       
$time_mid = bin2hex(substr($pr_bits,4, 2));
       
$time_hi_and_version = bin2hex(substr($pr_bits,6, 2));
       
$clock_seq_hi_and_reserved = bin2hex(substr($pr_bits,8, 2));
       
$node = bin2hex(substr($pr_bits,10, 6));
      
       
/**
         * Set the four most significant bits (bits 12 through 15) of the
         * time_hi_and_version field to the 4-bit version number from
         * Section 4.1.3.
         * @see http://tools.ietf.org/html/rfc4122#section-4.1.3
         */
       
$time_hi_and_version = hexdec($time_hi_and_version);
       
$time_hi_and_version = $time_hi_and_version >> 4;
       
$time_hi_and_version = $time_hi_and_version | 0x4000;
      
       
/**
         * Set the two most significant bits (bits 6 and 7) of the
         * clock_seq_hi_and_reserved to zero and one, respectively.
         */
       
$clock_seq_hi_and_reserved = hexdec($clock_seq_hi_and_reserved);
       
$clock_seq_hi_and_reserved = $clock_seq_hi_and_reserved >> 2;
       
$clock_seq_hi_and_reserved = $clock_seq_hi_and_reserved | 0x8000;
      
        return
sprintf('%08s-%04s-%04x-%04x-%012s',
           
$time_low, $time_mid, $time_hi_and_version, $clock_seq_hi_and_reserved, $node);
    }
david at ramaboo dot com 05-Jan-2009 12:51
Another UUID function. This time using /dev/random
<?php
/**
     * @brief Generates a Universally Unique IDentifier, version 4.
     *
     * This function generates a truly random UUID. The built in CakePHP String::uuid() function
     * is not cryptographically secure. You should uses this function instead.
     *
     * @see http://tools.ietf.org/html/rfc4122#section-4.4
     * @see http://en.wikipedia.org/wiki/UUID
     * @return string A UUID, made up of 32 hex digits and 4 hyphens.
     */
   
public function uuidSecure() {
       
       
$pr_bits = null;
       
$fp = @fopen('/dev/urandom','rb');
        if (
$fp !== false) {
           
$pr_bits .= @fread($fp, 16);
            @
fclose($fp);
        } else {
           
$this->cakeError('randomNumber');
        }
       
       
$time_low = bin2hex(substr($pr_bits,0, 4));
       
$time_mid = bin2hex(substr($pr_bits,4, 2));
       
$time_hi_and_version = bin2hex(substr($pr_bits,6, 2));
       
$clock_seq_hi_and_reserved = bin2hex(substr($pr_bits,8, 2));
       
$node = bin2hex(substr($pr_bits,10, 6));
       
       
/**
         * Set the four most significant bits (bits 12 through 15) of the
         * time_hi_and_version field to the 4-bit version number from
         * Section 4.1.3.
         * @see http://tools.ietf.org/html/rfc4122#section-4.1.3
         */
       
$time_hi_and_version = hexdec($time_hi_and_version);
       
$time_hi_and_version = $time_hi_and_version >> 4;
       
$time_hi_and_version = $time_hi_and_version | 0x4000;
       
       
/**
         * Set the two most significant bits (bits 6 and 7) of the
         * clock_seq_hi_and_reserved to zero and one, respectively.
         */
       
$clock_seq_hi_and_reserved = hexdec($clock_seq_hi_and_reserved);
       
$clock_seq_hi_and_reserved = $clock_seq_hi_and_reserved >> 2;
       
$clock_seq_hi_and_reserved = $clock_seq_hi_and_reserved | 0x8000;
       
        return
sprintf('%08s-%04s-%04x-%04x-%012s',
           
$time_low, $time_mid, $time_hi_and_version, $clock_seq_hi_and_reserved, $node);
    }
?>
mark at whytewaters dot com 25-Feb-2008 06:39
If you want many ids and performance of this function is an issue why not pull uniquid() out of the loop, eg:

$base = uniqueid();
$ids[] = array();

for ($index = 0; $index < 100000; $index++) 
    $ids[] = $base . '.' . $index;
nodkz at mail dot ru 15-Jan-2008 02:00
I use such UUID (it not RFC!!!)
(server_id)-(clientIP)-(unixtime)-(milliseconds)-(random)

 I can easyly determine which server at which time and who initiate creating of object.

<?php

$u
=uuid();   // 0001-7f000001-478c8000-4801-47242987
echo $u;
echo
"<br>";
print_r(uuidDecode($u)); // Array ( [serverID] => 0001 [ip] => 127.0.0.1 [unixtime] => 1200390144 [micro] => 0.28126525878906 )

function uuid($serverID=1)
{
   
$t=explode(" ",microtime());
    return
sprintf( '%04x-%08s-%08s-%04s-%04x%04x',
       
$serverID,
       
clientIPToHex(),
       
substr("00000000".dechex($t[1]),-8),   // get 8HEX of unixtime
       
substr("0000".dechex(round($t[0]*65536)),-4), // get 4HEX of microtime
       
mt_rand(0,0xffff), mt_rand(0,0xffff));
}

function
uuidDecode($uuid) {
   
$rez=Array();
   
$u=explode("-",$uuid);
    if(
is_array($u)&&count($u)==5) {
       
$rez=Array(
           
'serverID'=>$u[0],
           
'ip'=>clientIPFromHex($u[1]),
           
'unixtime'=>hexdec($u[2]),
           
'micro'=>(hexdec($u[3])/65536)
        );
    }
    return
$rez;
}

function
clientIPToHex($ip="") {
   
$hex="";
    if(
$ip=="") $ip=getEnv("REMOTE_ADDR");
   
$part=explode('.', $ip);
    for (
$i=0; $i<=count($part)-1; $i++) {
       
$hex.=substr("0".dechex($part[$i]),-2);
    }
    return
$hex;
}

function
clientIPFromHex($hex) {
   
$ip="";
    if(
strlen($hex)==8) {
       
$ip.=hexdec(substr($hex,0,2)).".";
       
$ip.=hexdec(substr($hex,2,2)).".";
       
$ip.=hexdec(substr($hex,4,2)).".";
       
$ip.=hexdec(substr($hex,6,2));
    }
    return
$ip;
}

?>
ken at smallboxsoftware 17-May-2007 09:34
Just to note this function is fairly slow, and can bring your script to a crawl if it is in a loop. Strangely if you run it as uniqid('', true) it runs much more quickly
Jason 27-Mar-2007 12:10
Neither the pseudo-random number rand() nor the Mersenne Twister algorithms are cryptographically strong, and this is well known.  Simply combining non-cryptographically strong algorithms doesn't not make a cryptographically strong algorithm either.  Mersenne Twister is a fast algorithm with good k-distribution which will give you numbers for a long time before it repeats itself.  MT, rand(), and MD5 should NOT be used for encryption, or for cookies that that store a session ID that gives personal information.  A simple application where non-collision of session IDs is highly preferred but not critical, such as storing a user's shopping cart items for when they return to your site (but not their personal information), IS a good use for the MT, rand() MD5, uniqid() and combinations thereof.
mailrinke at _cutthis_yahoo dot com 19-Feb-2007 05:06
I have been using mimecs version lately and do not think it's safe to think the results are always unqiue.

Although it could be just my bad programming, I found exactly 1 collission while debugging my code. It seems to me that if my code was incorrect it would have happened more than once.

I recommend anyone to include time as a factor of such an ID as to be a little more certain it is in fact unique.
lance_rushing at hotmail dot com 23-Jan-2007 12:43
wooshoofoo, the reason mimec is calling mt_rand multiple times is because the largest number mt_rand can produce is 2^31 (2147483647, as reported by mt_getrandmax() on my server).  RFC 4122 requires a 128 bit value.

Also they are not "4 digit sequeces", but 4 digit hexadecimal numbers.  16^4 == 2^16.

mimec's limiting each random result to 2^16 avoids problem of PHP's 2^32 integer max (http://php.net/manual/en/language.types.integer.php).

If you want to call mt_rand fewer times:  mimec's version calls mt_rand 8 times ( 16 bits * 8 = 128 bits ).  You *could* call mt_rand 5 times ( 31 bits + 31 bits + 31 bits + 31 bits + 4 bits = 128 bits ).  But then you would have keep all your values as strings.

Something like:

<?php
/**
 * Another (ugly) "random or pseudo-random" version of RFC 4122
 *
 * This version calls mt_rand() the fewest possible times.
 * if mt_getrandmax() == 2^31 then this will call mt_rand() 5 times YMMV
 *
 * Personally, I would use mimec's version
 * To handle the large values, we'll keep everything as strings.
 *
 * @return string
 */
function uuid() {   

    
// Generate 128 bit random sequence
    
$randmax_bits = strlen(base_convert(mt_getrandmax(), 10, 2));  // how many bits is mt_getrandmax()
    
$x = '';
     while (
strlen($x) < 128) {
        
$maxbits = (128 - strlen($x) < $randmax_bits) ? 128 - strlen($x) :  $randmax_bits;
        
$x .= str_pad(base_convert(mt_rand(0, pow(2,$maxbits)), 10, 2), $maxbits, "0", STR_PAD_LEFT);
     }

    
// break into fields
    
$a = array();
    
$a['time_low_part'] = substr($x, 0, 32);
    
$a['time_mid'] = substr($x, 32, 16);
    
$a['time_hi_and_version'] = substr($x, 48, 16);
    
$a['clock_seq'] = substr($x, 64, 16);
    
$a['node_part'] =  substr($x, 80, 48);
    
    
// Apply bit masks for "random or pseudo-random" version per RFC
    
$a['time_hi_and_version'] = substr_replace($a['time_hi_and_version'], '0100', 0, 4);
    
$a['clock_seq'] = substr_replace($a['clock_seq'], '10', 0, 2);

   
// Format output
   
return sprintf('%s-%s-%s-%s-%s',
       
str_pad(base_convert($a['time_low_part'], 2, 16), 8, "0", STR_PAD_LEFT),
       
str_pad(base_convert($a['time_mid'], 2, 16), 4, "0", STR_PAD_LEFT),
       
str_pad(base_convert($a['time_hi_and_version'], 2, 16), 4, "0", STR_PAD_LEFT),
       
str_pad(base_convert($a['clock_seq'], 2, 16), 4, "0", STR_PAD_LEFT),
       
str_pad(base_convert($a['node_part'], 2, 16), 12, "0", STR_PAD_LEFT));
}

?>

However, I think mimec's version is much more elegant.
wooshoofoo 06-Dec-2006 06:00
I'm not sure the previous function by mimec is really all that random.  For one thing, generating 8 small random 4 digit sequeces != generating one 32 digit sequence.