serialize

allan666 at gmail.com (2013-03-07 13:28:34)

Oddly, if you serialize a class that was previously unserialized, the class of the variable changes to string... Example:
$R = unserialize($serialized_object);
$R->method(); // this is ok
$str = serialize($R);
echo(get_class($R));
this will output "string"!!!!! whereas if the first line was
$R = new my_class();
it would output "my_class"!
I don't know if that is a bug, but the manual is not clear about that! (somehow $R in serialize($R) is being passed by reference, since it changes class).

nh at ngin dot de (2013-01-21 21:59:40)

Serializing floating point numbers leads to weird precision offset errors:

<?php

echo round(96.670000000000002, 2);
// 96.67

echo serialize(round(96.670000000000002, 2));
// d:96.670000000000002;

echo serialize(96.67);
// d:96.670000000000002;

?>

Not only is this wrong, but it adds a lot of unnecessary bulk to serialized data. Probably better to use json_encode() instead (which apparently is faster than serialize(), anyway).

Andrew B (2012-09-05 09:42:02)

When you serialize an array the internal pointer will not be preserved. Apparently this is the expected behavior but was a bit of a gotcha moment for me. Copy and paste example below.

<?php
//Internal Pointer will be 2 once variables have been assigned.
$array = array();
$array[] = 1;
$array[] = 2;
$array[] = 3;

//Unset variables. Internal pointer will still be at 2.     
unset($array[0]);
unset($array[1]);
unset($array[2]);

//Serialize
$serializeArray = serialize($array);

//Unserialize
$array = unserialize($serializeArray);

//Add a new element to the array
//If the internal pointer was preserved, the new array key should be 3.
//Instead the internal pointer has been reset, and the new array key is 0.
$array[] = 4;

//Expected Key - 3
//Actual Key - 0
echo "<pre>" , print_r($array, 1) , "</pre>";
?>

Anonymous (2012-02-28 19:44:54)

Please! please! please! DO NOT serialize data and place it into your database. Serialize can be used that way, but that's missing the point of a relational database and the datatypes inherent in your database engine. Doing this makes data in your database non-portable, difficult to read, and can complicate queries. If you want your application to be portable to other languages, like let's say you find that you want to use Java for some portion of your app that it makes sense to use Java in, serialization will become a pain in the buttocks. You should always be able to query and modify data in the database without using a third party intermediary tool to manipulate data to be inserted.
I've encountered this too many times in my career, it makes for difficult to maintain code, code with portability issues, and data that is it more difficult to migrate to other RDMS systems, new schema, etc. It also has the added disadvantage of making it messy to search your database based on one of the fields that you've serialized.
That's not to say serialize() is useless. It's not... A good place to use it may be a cache file that contains the result of a data intensive operation, for instance. There are tons of others... Just don't abuse serialize because the next guy who comes along will have a maintenance or migration nightmare.

pgl at yoyo dot org (2011-02-03 09:13:28)

Another suggestion for coping with binary data in serialize()d variables is just to base64_encode() those fields before serializing. It will increase the size of the variable, but not by too much.

travis at travishegner dot com (2010-03-02 12:37:31)

If serializing objects to be stored into a postgresql database, the 'null byte' injected for private and protected members throws a wrench into the system. Even pg_escape_bytea() on the value, and storing the value as a binary type fails under certain circumstances.

For a dirty work around:
<?php

$serialized_object = serialize($my_object);
$safe_object = str_replace("\0", "~~NULL_BYTE~~", $serialized_object);

?>

this allows you to store the object in a readable text format as well. When reading the data back:

<?php

$serialized_object = str_replace("~~NULL_BYTE~~", "\0", $safe_object);
$my_object = unserialize($serialized_object);

?>

The only gotcha's with this method is if your object member names or values may somehow contain the odd "~~NULL_BYTE~~" string. If that is the case, then str_replace() to a string that you are guaranteed not to have any where else in the string that serialize() returns.
Also remember to define the class before calling unserialize().

If you are storing session data into a postgresql database, then this workaround is an absolute must, because the $data passed to the session's write function is already serialized.

Thanks,
Travis Hegner

Anonymous (2010-01-21 09:23:44)

you should really use mysql_real_escape_string() for escaping (serialized) strings that got thrown into a query (visit php.net/mysql_real_escape_string for further information)

donotmail at me dot com (2009-09-02 11:16:57)

NOTE: php's serialize does not properly serialize arrays with which a slice of the array is a reference to the array itself, observe:

<?php
$a = array();
$a[0] = "blah";
$a[1] =& $a;

$a[1][0] = "pleh"; // $a[0] === "pleh"

$b = unserialize(serialize($a));

// $b[0] == "pleh", $b[1][0] == "pleh"

$b[1][0] = "blah";
?>

now $b[1][0] == "blah", but $b[0] == "pleh"
after serializing and unserializing, slice 1 is no longer a reference to the array itself... I have found no way around this problem... even manually modifying the serialized string from 
'a:2:{i:0;s:4:"pleh";i:1;a:2:{i:0;s:4:"pleh";i:1;R:3;}}'
to
'a:2:{i:0;s:4:"pleh";i:1;R:1;}'

to force the second slice to be a reference to the first element of the serialization (the array itself), it seemed to work at first glance, but then unreferences it when you alter it again, observe:

<?php
    $testser = 'a:2:{i:0;s:4:"pleh";i:1;R:1;}';

    $tmp = unserialize($testser);

    print_r($tmp);

    print "\n-----------------------\n";

    $tmp[1][0] = "blah";

    print_r($tmp);
 
?>

outputs:
Array
(
    [0] => pleh
    [1] => Array
 *RECURSION*
)

-----------------------
Array
(
    [0] => pleh
    [1] => Array
        (
            [0] => blah
            [1] => Array
                (
                    [0] => pleh
                    [1] => Array
 *RECURSION*
                )

        )

)

Alessandro Segala (2009-07-17 08:18:58)

I needed to serialize an array to store it inside a database.
I was looking for a fast, simple way to do serialization, and I came out with 2 options: serialize() or json_encode().
I ran some benchmarks to see which is the faster, and, surprisingly, I found that serialize() is always between 46% and 96% SLOWER than json_encode().
So, if you don't need to serialize objects and have the json extension available, consider using it instead of serialize().

NWdev (2009-06-08 19:27:28)

If you're serializing a small array of values and sending them from page to page for display/navigation purposes (versus storing them as session values) via $_GET, save yourself some time and ensure your php.ini magic_quotes_gpc values are set the same on your hosting and localhost severs.
In my case I had magic_quotes_gpc OFF locally, and it took a while to see why the values weren't unserializing properly (actually at all -- unserialize produced FALSE) on the action page (form action page) of the hosted version. A quick look at phpinfo() showed a difference -- Thanks to others mentioning use of magic_quotes when saving to mysql triggered a thot that this mismatch might cause the problem.

barbuj at NOSPAM dot yahoo dot com (2007-11-14 14:19:21)

In my specific situation, I wanted to be able to pass some data from page to page, but without relying on a session variable. The answer I came up with was to serialize() the object in question, pass it on to the next page as a hidden form field, then unserialize() it. When ALL class variables are public, this worked fine. However, if there was at least one private/protected variable, the code no longer worked as expected ("Fatal error: Call to a member function display() on a non-object in page2.php on line 4")

As others have already mentioned, private/protected class variables will not behave nicely (private variables are prefixed by class_name + �, while protected variables are only prefixed by � - when looking at the page source using Firefox). Internet Explorer does NOT display the extra character between the class name and variable name, but the code still doesn't work as one would expect.

Suppose you have a simple class:

testclass.php
=============
<?php
class TestClass {
  var $i = 1;

  function display() {
    echo "i=" . $this->i;
  }
?>

page1.php
=========
<?php
  require_once 'testclass.php';
  $tc = new TestClass;
  $tc->display();
?>
<form method = "post" action = "page2.php">
<input type = "hidden" name = "str" value = "<?php echo htmlspecialchars( serialize( $tc ) ); ?>">
<input type = "submit">
</form>

page2.php
=========
<?php
  require_once 'testclass.php';
  $tc = unserialize( stripslashes( htmlspecialchars_decode( $_POST["str"] ) ) );
  $tc->display();
?>

The fix, suggested by evulish on #php/irc.dal.net, is to replace htmlspecialchars()/htmlspecialchars_decode() by base64_encode()/base64_decode. The code becomes:

page1.php
=========
<input type = "hidden" name = "str" value = "<?php echo base64_encode( serialize( $tc ) ); ?>">

page2.php
=========
<?php
  $tc = unserialize( base64_decode( $_POST["str"] ) );
?>

Hope this will help someone...

Jeex (2007-11-06 02:59:07)

It may be worth noting that, depending on the size of the object you are serializing, the serialize function can take up a lot of memory.
If your script isn't working as expected, your serialize call may have pushed the memory usage over the limit set by memory_limit in php.ini.
More info on memory limits here: http://www.php.net/manual/en/ini.core.php

friday13 at ig dot com dot br (2007-06-19 07:15:59)

I have problem to use serialize function with hidden form field and the resolution was use htmlentities.

Ex.:

<?

$lista = array( 'pera', 'ma?a', 'laranja' );

print "< input type='hidden' name='teste' value='htmlentities( serialize( $lista ) )'" >";

?>

Ates Goral (2006-07-18 11:59:01)

Corrections/clarifications to "Anatomy of a serialize()'ed value":
All strings appear inside quotes. This applies to string values, object class names and array key names. For example:
s:3:"foo"
O:7:"MyClass":1:{...
a:2:{s:3:"bar";i:42;...
Object property names and values are delimited by semi-colons, not colons. For example:
O:7:"MyClass":2:{s:3:"foo";i:10;s:3:"bar";i:20}
Double/float values are represented as:
d:0.23241446

egingell at sisna dot com (2006-05-15 17:12:29)

<?
/*
Anatomy of a serialize()'ed value:

 String
 s:size:value;

 Integer
 i:value;

 Boolean
 b:value; (does not store "true" or "false", does store '1' or '0')

 Null
 N;

 Array
 a:size:{key definition;value definition;(repeated per element)}

 Object
 O:strlen(object name):object name:object size:{s:strlen(property name):property name:property definition;(repeated per property)}

 String values are always in double quotes
 Array keys are always integers or strings
    "null => 'value'" equates to 's:0:"";s:5:"value";',
    "true => 'value'" equates to 'i:1;s:5:"value";',
    "false => 'value'" equates to 'i:0;s:5:"value";',
    "array(whatever the contents) => 'value'" equates to an "illegal offset type" warning because you can't use an
    array as a key; however, if you use a variable containing an array as a key, it will equate to 's:5:"Array";s:5:"value";',
     and
    attempting to use an object as a key will result in the same behavior as using an array will.
*/
?>

MC_Gurk at gmx dot net (2006-01-03 07:44:56)

If you are going to serialie an object which contains references to other objects you want to serialize some time later, these references will be lost when the object is unserialized.
The references can only be kept if all of your objects are serialized at once.
That means:
$a = new ClassA();
$b = new ClassB($a); //$b containes a reference to $a;
$s1=serialize($a);
$s2=serialize($b);
$a=unserialize($s1);
$b=unserialize($s2);
now b references to an object of ClassA which is not $a. $a is another object of Class A.
use this:
$buf[0]=$a;
$buf[1]=$b;
$s=serialize($buf);
$buf=unserialize($s);
$a=$buf[0];
$b=$buf[1];
all references are intact.

Alexander Podgorny (2005-12-02 23:47:44)

Here is an example of a base class to implement object persistence using serialize and unserialize:

<?php
class Persistent
{
    var $filename;
        
    /**********************/
    function Persistent($filename)
    {
        $this->filename = $filename;
        if(!file_exists($this->filename)) $this->save();
    }
    /**********************/
    function save()
    {
        if($f = @fopen($this->filename,"w"))
        {
            if(@fwrite($f,serialize(get_object_vars($this))))
            {
                @fclose($f);
            }
            else die("Could not write to file ".$this->filename." at Persistant::save");
        }
        else die("Could not open file ".$this->filename." for writing, at Persistant::save");
        
    }
    /**********************/
    function open()
    {
        $vars = unserialize(file_get_contents($this->filename));
        foreach($vars as $key=>$val)
        {            
            eval("$"."this->$key = $"."vars['"."$key'];");
        }
    }
    /**********************/
}

?>

When an object is extended from this one it can be easily saved and re-opened using it's own methods as follows:

<?

class foo extends Persistent
{
   var $counter;
   function inc()
   {
       $this->counter++;
   }
}

$fooObj = new $foo;
$foo->open();
print $foo->counter; // displays incrementing integer as page reloads
$foo->inc();
$foo->save();

?>

stephen dot adamson at smius dot com (2005-11-01 09:40:11)

If you are serializing an object to store it in the database, using __sleep() can save you some space. The following code will not store empty (null) variables in the serialized string. For my purposes this saved a lot of space, since some of the member variables would not be given values.
function __sleep()
{
$allVars = get_object_vars($this);
$toReturn = array();
foreach(array_keys($allVars) as $name)
{
if (isset($this->$name))
{
$toReturn[] = $name;
}
}
return $toReturn;
}

wiart at yahoo dot com (2004-12-16 03:40:40)

Warning: on 64 bits machines, if you use a long string only composed of numbers as a key in an array and serialize/unserialize it, you can run into problems:
an example code:
$arr["20041001103319"] = 'test';
var_dump( $arr);
$arr_in_str = serialize($arr);
print "Now result is: $arr_in_str<BR />";
$final_arr = unserialize($arr_in_str);
print "The final unserialized array:<BR />";
var_dump($final_arr);
The result:
array(1) { [20041001103319]=> string(4) "test" }
Now result is: a:1:{i:20041001103319;s:4:"test";}
The final unserialized array:
array(1) { [683700183]=> string(4) "test" }
As you can see, the original array :
$arr["20041001103319"] = "test"
after serialize/unserialize is:
$arr[683700183] = "test"
As you can see, the key has changed ...
Apparently a problem of implicit casting + integer overflow (I posted a PHP bug report: http://bugs.php.net/bug.php?id=31117)
I tested with the latest 4.3.10 compiled on my laptop (32 bits, Mandrake 9.1) --> no such problem. But compiled on AMD 64 bits (Red Hat Taroon), the problem is present.
Hope this will help some of you to not loose almost a whole day of debugging ;-)

paul at moveNOtoSPAMiceland DOT com (2004-11-18 08:29:42)

I was trying to submit a serialized array through a hidden form field using POST and was having a lot of trouble with the quotes. I couldn't figure out a way to escape the quotes in the string so that they'd show up right inside the form, so only the characters up to the first set of quotes were being sent.
My solution was to base64_encode() the string, put that in the hidden form field, and send that through the POST method. Then I decoded it (using base64_decode()) on the other end. This seemed to solve the problem.

brooklynphil hotmail com (2004-10-06 13:07:40)

In my experience of the (far and few in between) times when it is necessary to store large (megabytes) amounts of data in a php-readable ascii file format, using var_export() and require() is much faster serialize() and unserialize(file_get_contents()) (I have tested it on classes encapsulating arrays, i.e. <?php class blah { $a = array(megsofstuffhere); $b = array(megsofstuffhere); $c=... }; $data = new blah ?>)   (var_export is also more readable and notepadable than serialize!)

However, since var_export produces only the rvalue (or uninstantiation), of the data needed, dont forget to cosmetically modify output of var_export( , true): in case of classes, you will need to use "$var = new classname" (and rename a possible stdClass) and in case of arrays, etc, just use "$var = ".var_export().  The file being var_export()ed to and require()ed must also be encapculated in "<?php ?>" when written to (or else its just dumped to browser/stdout as a textfile).

Hope this helps :)

--Phil

P.S. (in cases of small amounts of data, serialize is better, especially for cross-URL data transport using <?php base64_encode(gzdeflate(serialize()))?>)

hfuecks at phppatterns dot com (2004-04-21 20:20:50)

Regarding serializing PHP data types to Javascript, following Ivans note below, theres an example at http://www.tekool.net/php/js_serializer/.
The basic serialization looks good although, in its current form, it works on the basis of generating Javascript source which a browser executes as a page loads. Using Javascripts eval() the same can be done with strings containing Javascript if youre working with something like XMLHTTPRequest

pli9 at itsa dot ucsf dot edu (2003-01-31 23:05:05)

I have also written some code for importing serialized PHP data into PERL and then writing it back into PHP. I think the similar library posted above is actually more robust for a few select cases, but mine is more compact and a little easier to follow. I'd really like comments if anyone finds this useful or has improvements. Please credit me if you use my code.
http://www.hcs.harvard.edu/~pli/code/serialPHP.pm

MarkRoedel at letu dot edu (2000-02-21 22:58:10)

A call to serialize() appears to mess with the array's internal pointer. If you're going to be walking through your array after serializing it, you'll want to make a call to reset() first.