time_sleep_until
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()); }
uniqid
(PHP 4, PHP 5)
uniqid — Gera um ID unico
Descrição
string uniqid
( string $prefix
[, bool $lcg
] )
uniqid() retorna um identificador unico prefixado baseado no tempo atual em milionésimos de segundo. O Prefixo pode ser usado se você gera identificadores em vários servidores simultaneamente pode acontecer de gerar o identificador no mesmo milionésimo de segundo. Prefix pode ter até 114 caracteres.
Se o parâmetro lcg , que é opcional, for TRUE, uniqid() irá adicionar a entropia "LCG combinada" ao final do valor retornado, o que deve fazer o resultado mais unico.
Com o parâmetro prefix vazio, a string retornada terá 13 caracteres. Se o parâmetro lcg for TRUE, terá 23 caracteres.
Nota: O parâmetro lcg somente esta disponivel no PHP 4 e PHP 3.0.13 e posterior.
Se você precisa um identificador unico e pretende dar o seu identificador via rede(ex. cookies de seção), é recomendado que você use algo como
<?php
// sem prefixo
$token = md5(uniqid(""));
// melhor, dificil de adivinhar
$better_token = md5(uniqid(rand(), true));
?>
Isto irá criar um identificador de 32 caracteres(a 128 bit hex number) que é extremamente dificil de prever.
add a note
User Contributed Notesuniqid
John Haugeland from FullOf.BS
26-May-2009 03:11
26-May-2009 03:11
Marius Karthaus
23-Jan-2009 05:10
23-Jan-2009 05:10
Better yet, just use the php5-uuid extension and this class to create 'official' UUIDs at high speed (On my system 1000 UUIDs are created in 0.0064 seconds)
<?php
Usage:
$uuid=new uuid();
echo $uuid->v1();
class uuid {
/**
* This class enables you to get real uuids using the OSSP library.
* Note you need php-uuid installed.
* On my system 1000 UUIDs are created in 0.0064 seconds.
*
* @author Marius Karthaus
*
*/
protected $uuidobject;
/**
* On long running deamons i've seen a lost resource. This checks the resource and creates it if needed.
*
*/
protected function create() {
if (! is_resource ( $this->uuidobject )) {
uuid_create ( &$this->uuidobject );
}
}
/**
* Return a type 1 (MAC address and time based) uuid
*
* @return String
*/
public function v1() {
$this->create ();
uuid_make ( $this->uuidobject, UUID_MAKE_V1 );
uuid_export ( $this->uuidobject, UUID_FMT_STR, &$uuidstring );
return trim ( $uuidstring );
}
/**
* Return a type 4 (random) uuid
*
* @return String
*/
public function v4() {
$this->create ();
uuid_make ( $this->uuidobject, UUID_MAKE_V4 );
uuid_export ( $this->uuidobject, UUID_FMT_STR, &$uuidstring );
return trim ( $uuidstring );
}
/**
* Return a type 5 (SHA-1 hash) uuid
*
* @return String
*/
public function v5() {
$this->create ();
uuid_make ( $this->uuidobject, UUID_MAKE_V5 );
uuid_export ( $this->uuidobject, UUID_FMT_STR, &$uuidstring );
return trim ( $uuidstring );
}
}
?>
Marius Karthaus
22-Jan-2009 11:39
22-Jan-2009 11:39
I've created a class from the function below. With it, creating a lot of uuids beomes about 100 times faster because you do not need to fopen() for each uuid.
uuid::get() still works as a static function if you just need a single uuid.
<?php
class uuid {
protected $urand;
public function __construct() {
$this->urand = @fopen ( '/dev/urandom', 'rb' );
}
/**
* @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 get() {
$pr_bits = false;
if (is_a ( $this, 'uuid' )) {
if (is_resource ( $this->urand )) {
$pr_bits .= @fread ( $this->urand, 16 );
}
}
if (! $pr_bits) {
$fp = @fopen ( '/dev/urandom', 'rb' );
if ($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 );
}
}
sean at seancolombo dot com
06-Jan-2009 01:56
06-Jan-2009 01:56
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 06:51
05-Jan-2009 06:51
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);
}
?>
valerylourie at gmail dot com
13-Aug-2008 06:38
13-Aug-2008 06:38
To dholmes at cfdsoftware dot net:
sprintf parameter should be changed to '%04x%04x-%04x-4%03x-%04x-%04x%04x%04x' - 4 in GUID V4 should be first, but not last digit in the third group
lucas at paivlek dot net
12-Jul-2008 02:07
12-Jul-2008 02:07
Hi all,
I'm not sure what performance problems you guys are talking about, but I benchmarked 10000 of each of the functions outlined in this comments section alongside the uniqid method. Where function names are duplicated (because everyone likes to call their method uuid), i've added part of that contributer's name to the function name.
So if you want something unique and you think that uniqid() is slowing your scripts down, maybe rand() is your only answer :)
See the output (quickest to slowest):
Array
(
[0] => Array
(
[strMethod] => rand
[intRepetitions] => 10000
[flTotalTime] => 0.104878902435
[flAverageTime] => 1.04878902435E-5
[flShortestTime] => 3.81469726562E-6
[flLongestTime] => 0.000297069549561
[exampleOutput] => 19689
)
[1] => Array
(
[strMethod] => uniqid
[intRepetitions] => 10000
[flTotalTime] => 0.123727083206
[flAverageTime] => 1.23727083206E-5
[flShortestTime] => 3.81469726562E-6
[flLongestTime] => 9.3936920166E-5
[exampleOutput] => 48782b3c3e997
)
[2] => Array
(
[strMethod] => uniqid2
[intRepetitions] => 10000
[flTotalTime] => 0.132393836975
[flAverageTime] => 1.32393836975E-5
[flShortestTime] => 3.81469726562E-6
[flLongestTime] => 0.000135898590088
[exampleOutput] => 48782b3d282c
)
[3] => Array
(
[strMethod] => uuid_mimec
[intRepetitions] => 10000
[flTotalTime] => 0.192065000534
[flAverageTime] => 1.92065000534E-5
[flShortestTime] => 3.81469726562E-6
[flLongestTime] => 0.000221014022827
[exampleOutput] => fc1b24b7-59e0-4935-88f2-06a6fde922d8
)
[4] => Array
(
[strMethod] => uuid_maciej
[intRepetitions] => 10000
[flTotalTime] => 0.219671964645
[flAverageTime] => 2.19671964645E-5
[flShortestTime] => 3.81469726562E-6
[flLongestTime] => 0.000162839889526
[exampleOutput] => 4c55bc99-b5bc-0f88-0387-00001a51bd33
)
[5] => Array
(
[strMethod] => uuid_dholmes
[intRepetitions] => 10000
[flTotalTime] => 0.221670866013
[flAverageTime] => 2.21670866013E-5
[flShortestTime] => 3.81469726562E-6
[flLongestTime] => 0.000198125839233
[exampleOutput] => cd79ac78-af08-a804-6567-3c4940e3252b
)
[6] => Array
(
[strMethod] => newid
[intRepetitions] => 10000
[flTotalTime] => 0.221807956696
[flAverageTime] => 2.21807956696E-5
[flShortestTime] => 3.81469726562E-6
[flLongestTime] => 0.000211954116821
[exampleOutput] => user_9e2e305b4c4340397617baf285f11eb348782b3d0354b3.
095706961215834941127.0.0.1
)
[7] => Array
(
[strMethod] => uuid_nodkz
[intRepetitions] => 10000
[flTotalTime] => 0.477306842804
[flAverageTime] => 4.77306842804E-5
[flShortestTime] => 3.81469726562E-6
[flLongestTime] => 0.000205039978027
[exampleOutput] => 0001-7f000001-48782b3e-0952-ef8e7ec8
)
[8] => Array
(
[strMethod] => uuid_lance
[intRepetitions] => 10000
[flTotalTime] => 0.668512821198
[flAverageTime] => 6.68512821198E-5
[flShortestTime] => 3.81469726562E-6
[flLongestTime] => 0.00036096572876
[exampleOutput] => fa465717-26bf-4af9-987f-33bb54905678
)
)
mark at whytewaters dot com
26-Feb-2008 12:39
26-Feb-2008 12:39
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 08:00
15-Jan-2008 08:00
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 $ip;
}
?>
smp_info at yahoo dot com
06-Sep-2007 10:38
06-Sep-2007 10:38
This function is painfully slow if you're using it to give images random names inside of a loop. The following function will give you a random name *every* time and is much faster.
<?php
function nameImage($imgExtension)
{
return time() . substr(md5(microtime()), 0, rand(5, 12)) . $imgExtension;
}
?>
dot dot dot dot dot alexander at gmail dot com
13-Jun-2007 07:28
13-Jun-2007 07:28
I use this mangle currently:
( inserts the IP, uses time() and a prefix, aside the uniqid)
<?php
if(!function_exists("newid")){
function newid($prefix = "user_"){
return ($prefix . uniqid( hash("md5", time()), TRUE ) . time() . @$_SERVER['REMOTE_ADDR']);
}/* endfunction */
}/* endif */
?>
ken at smallboxsoftware
17-May-2007 02:34
17-May-2007 02:34
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
rjchallen at gmail dot com
17-May-2007 01:08
17-May-2007 01:08
If you can guarantee a connection to mysql when you need your UUID then you can wrap up MySQL's (v5+) function.
function uuid() {
return mysql_result(mysql_query('Select UUID()'),0);
}
kristoffer dot paro at gmail dot com
13-May-2007 01:05
13-May-2007 01:05
In response to the notes about UUID generation added by mimec and lance_rushing at hotmail dot com.
Calling mt_rand the fewest possible times is not necessarily the fastest, if it heavily utilizes string handling routines. I did a quick benchmark between the two functions and discovered that lance's function (using only 5 mt_rands) was about 6.5 times _slower_ than mimec's on my system.
Jason
27-Mar-2007 06:10
27-Mar-2007 06:10
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 11:06
19-Feb-2007 11:06
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.
Emery
31-Jan-2007 06:13
31-Jan-2007 06:13
The example given in this document for a "better token" should be:
<?php
$better_token = uniqid(md5(rand()), true);
?>
As it is now, the result isn't guaranteed to be unique, because MD5 has collisions.
lance_rushing at hotmail dot com
23-Jan-2007 06:43
23-Jan-2007 06:43
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.
mailbox1 at highhost dot net
23-Dec-2006 10:22
23-Dec-2006 10:22
Also you may use this if you like it.
function uniqid2() {
return dechex(time()).dechex(mt_rand(1,65535));
}
wooshoofoo
07-Dec-2006 12:00
07-Dec-2006 12:00
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.
mimec
25-Aug-2006 06:36
25-Aug-2006 06:36
Here is the correct version of a function generating a pseudo-random UUID according to RFC 4122:
<?php
function uuid()
{
return sprintf( '%04x%04x-%04x-%04x-%04x-%04x%04x%04x',
mt_rand( 0, 0xffff ), mt_rand( 0, 0xffff ), mt_rand( 0, 0xffff ),
mt_rand( 0, 0x0fff ) | 0x4000,
mt_rand( 0, 0x3fff ) | 0x8000,
mt_rand( 0, 0xffff ), mt_rand( 0, 0xffff ), mt_rand( 0, 0xffff ) );
}
?>
The version and variant is located at the MSB (most significant bits) of the time_hi_and_version and clock_seq_hi_and_reserved fields, not the LSB as in dholmes version.
admin at code-dynasty dot net
09-Jul-2006 01:46
09-Jul-2006 01:46
I'm not too fond of the recommendation to use an MD5 of the unique ID for session IDs. It would be a better idea just to use uniqueid(rand(), true) without the MD5, because even though it's a rare circumstance, MD5 is a hash, not an encryption, which means it has collisions. Therefore you theoretically could have multiple users given the same session ID which could result in one user's ability to access another user's data.
dholmes at cfdsoftware dot net
09-May-2006 01:26
09-May-2006 01:26
WARNING : I believe there are a couple of mistakes in the function provided just below by maciej dot strzelecki at gmail dot com. Namely, that in the two substr_replace() calls, the third parameters should respectively be 12 (instead of 11) and 6 (instead of 5).
Considering the importance of this function, I went to read RFC 4122 myself, and found the discrepancy. I therefore chose to write my own function, inspired by the previous one, but with a few enhancements detailed in the comments. On the downside, it might be slightly less easy to understand at first glance.
Please feel free to use it yourself. Thank you also in advance for any feedback at dholmes at cfdsoftware.net .
<?php
/**
* Generates a Universally Unique IDentifier, version 4.
*
* RFC 4122 (http://www.ietf.org/rfc/rfc4122.txt) defines a special type of Globally
* Unique IDentifiers (GUID), as well as several methods for producing them. One
* such method, described in section 4.4, is based on truly random or pseudo-random
* number generators, and is therefore implementable in a language like PHP.
*
* We choose to produce pseudo-random numbers with the Mersenne Twister, and to always
* limit single generated numbers to 16 bits (ie. the decimal value 65535). That is
* because, even on 32-bit systems, PHP's RAND_MAX will often be the maximum *signed*
* value, with only the equivalent of 31 significant bits. Producing two 16-bit random
* numbers to make up a 32-bit one is less efficient, but guarantees that all 32 bits
* are random.
*
* The algorithm for version 4 UUIDs (ie. those based on random number generators)
* states that all 128 bits separated into the various fields (32 bits, 16 bits, 16 bits,
* 8 bits and 8 bits, 48 bits) should be random, except : (a) the version number should
* be the last 4 bits in the 3rd field, and (b) bits 6 and 7 of the 4th field should
* be 01. We try to conform to that definition as efficiently as possible, generating
* smaller values where possible, and minimizing the number of base conversions.
*
* @copyright Copyright (c) CFD Labs, 2006. This function may be used freely for
* any purpose ; it is distributed without any form of warranty whatsoever.
* @author David Holmes <dholmes@cfdsoftware.net>
*
* @return string A UUID, made up of 32 hex digits and 4 hyphens.
*/
function uuid() {
// The field names refer to RFC 4122 section 4.1.2
return sprintf('%04x%04x-%04x-%03x4-%04x-%04x%04x%04x',
mt_rand(0, 65535), mt_rand(0, 65535), // 32 bits for "time_low"
mt_rand(0, 65535), // 16 bits for "time_mid"
mt_rand(0, 4095), // 12 bits before the 0100 of (version) 4 for "time_hi_and_version"
bindec(substr_replace(sprintf('%016b', mt_rand(0, 65535)), '01', 6, 2)),
// 8 bits, the last two of which (positions 6 and 7) are 01, for "clk_seq_hi_res"
// (hence, the 2nd hex digit after the 3rd hyphen can only be 1, 5, 9 or d)
// 8 bits for "clk_seq_low"
mt_rand(0, 65535), mt_rand(0, 65535), mt_rand(0, 65535) // 48 bits for "node"
);
}
?>
maciej dot strzelecki at gmail dot com
16-Apr-2006 04:09
16-Apr-2006 04:09
This is an implementation of version 4 UUID, which is generating UUIDs from truly-random numbers.
<?php
/* Copyright 2006 Maciej Strzelecki
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation; either version 2 of the License, or
(at your option) any later version.
This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program; if not, write to the Free Software
Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA */
function uuid()
{
// version 4 UUID
return sprintf(
'%08x-%04x-%04x-%02x%02x-%012x',
mt_rand(),
mt_rand(0, 65535),
bindec(substr_replace(
sprintf('%016b', mt_rand(0, 65535)), '0100', 11, 4)
),
bindec(substr_replace(sprintf('%08b', mt_rand(0, 255)), '01', 5, 2)),
mt_rand(0, 255),
mt_rand()
);
}
?>
03-Apr-2006 04:16
if you're generating UUIDs you should really check out http://www.ietf.org/rfc/rfc4122.txt first ('version 4' UUID, chapter 4.4), as they are not completely random.