Doc #78190 [Com]: mb_split now return false with ISO strings

Edit report at https://bugs.php.net/bug.php?id=78190&edit=1

 ID:                 78190
 Comment by:         Isaac1866Williams at gmail dot com
 Reported by:        mvachette at adequasys dot com
 Summary:            mb_split now return false with ISO strings
 Status:             Re-Opened
 Type:               Documentation Problem
 Package:            mbstring related
 Operating System:   Windows 10
 PHP Version:        7.2.19
 Block user comment: N
 Private report:     N

 New Comment:

Hello,

In the case of failure to perform preg_split due to invalid inputs, preg_split will return false.
Example-

$string = array("sdasds"); // Invalid string
$array=preg_split('/[\s,]+/',$string);
var_dump($array); //false

$string = "sdasds";
$array=preg_split('j',$string); // Invalid pattern
var_dump($array); //false

(https://www.mycenturahealth.me/)github.com


Previous Comments:
------------------------------------------------------------------------
[2019-06-20 15:47:19] [email protected]

Right, this should probably be documented. This change was backported in https://github.com/php/php-src/commit/0ecac37c40a27ffbd59f34b5920735ee0b7f994c#diff-3b307ca282b8a468c86220f768eea0d7 which is in 7.1.28, 7.2.18 and 7.3.5. For other mb_regex functions apart from mb_split and mb_ereg_match this was already checked previously, though I'm not sure since when.

------------------------------------------------------------------------
[2019-06-20 15:30:40] mvachette at adequasys dot com

Ok, I figure to clear the error using mb_regex_encoding, thanks for the suggestion.

But is there any documentation available about this change of behaviour? For information, I found this burried into Smarty lib. I can ensure correct behaviour of mb_split on my app, by maybe I will not be the lone to be impacted.

------------------------------------------------------------------------
[2019-06-20 15:21:27] [email protected]

The default encoding is UTF-8, use mb_regex_encoding() if you are working with something else.

------------------------------------------------------------------------
[2019-06-20 15:15:44] [email protected]

I cannot reproduce this, it seems to work pretty consistently: https://3v4l.org/QqYjC

------------------------------------------------------------------------
[2019-06-20 15:09:11] mvachette at adequasys dot com

Description:
------------
It seems that something changes in 7.2.18 when using mb_split on ISO string.

On at least 7.2.16 & 7.2.17, using this function with ISO encoding string containing accents works fine.

On a new env with version 7.2.18, the same code now return "false" (new undocumented behaviour)



Test script:
---------------
var_dump(mb_split('-', 'e-a'));
//works fine on any version

var_dump(mb_split('-', 'é-a'));
// works if string is UTF8 encoded
// return "false" with an ISO encoded string on PHP 7.2.18



------------------------------------------------------------------------



--
Edit this bug report at https://bugs.php.net/bug.php?id=78190&edit=1