to get length of binary string in bytes - tiny benchmark

get_binlen_bench.php

<?php
$count = 100;

$data = "あいうえお";

for ($i= 1; $i < 10; $i++) {
   $data .= $data;
}

echo phpversion(), PHP_EOL;
echo "func_overload ?", PHP_EOL;
var_dump(function_exists('mb_internal_encoding') &&
                    (((int)ini_get('mbstring.func_overload')) & 2));
echo "show target data length : ", (strlen(bin2hex($data)) / 2), PHP_EOL;

echo "******************** start strlen(bin2hex", PHP_EOL;
$start = microtime();
for ($i= 1; $i < $count; $i++) {
   strlen(bin2hex($data)) / 2;
}
$end = microtime();

var_dump($end - $start);

echo "******************** start Util::strlen", PHP_EOL;

$start = microtime();
for ($i= 1; $i < $count; $i++) {
   Util::strlen($data);
}
$end = microtime();

var_dump($end - $start);

echo "******************** start Util2::strlen", PHP_EOL;

$start = microtime();
for ($i= 1; $i < $count; $i++) {
   Util2::strlen($data);
}
$end = microtime();

var_dump($end - $start);


class Util
{

   static public function strlen($data)
   {
       if (function_exists('mb_internal_encoding') &&
           (((int)ini_get('mbstring.func_overload')) & 2)) {
           return mb_strlen($data, '8bit');
       } else {
           return strlen($data);
       }
   }
}


class Util2
{

   static public function strlen($data)
   {
       static $overload;
       if ($overload === null) {
           if (function_exists('mb_internal_encoding') &&
               (((int)ini_get('mbstring.func_overload')) & 2)) {
               $overload = true;
           } else {
               $overload = false;
           }
       }

       return ($overload) ? mb_strlen($data, '8bit') : strlen($data);
   }
}

log_php535ubuntu

5.3.5-1ubuntu7.7
func_overload ?
bool(false)
show target data length : 7680
******************** start strlen(bin2hex
float(0.004369)
******************** start Util::strlen
float(0.002455)
******************** start Util2::strlen
float(0.001262)

log_php540

5.4.0
func_overload ?
bool(false)
show target data length : 7680
******************** start strlen(bin2hex
float(0.0041910000000001)
******************** start Util::strlen
float(0.000552)
******************** start Util2::strlen
float(0.00010500000000002)

Why not simply:

static public function strlen($data) {
static $mbStrlen = function_exists('mb_strlen');
return $mbStrlen ? mb_strlen($str, '8bit') : strlen($str);
}


@marc-mabe

as I mentioned past, zendframework/zendframework#546,
above checking environment code is borrowed from Zend\OpenId's.

• OpenId's "strlen" is coding by Padraic, so I just following it blindly :-)

In addition, If checking only function_exists("mb_strlen"),
all mbstring loaded system not be enforced mb_strlen?
mb_strlen($data, '8bit') is cost than strlen.

And, mbstring.func_overload 's Changeable is PHP_INI_SYSTEM | PHP_INI_PERDIR.
http://php.net/manual/en/mbstring.configuration.php
when checking system strictly,
it can be memorize to static as enforce using mb_strlen($data, '8bit').
(mb_strlen function can define as user function
https://github.com/cakephp/cakephp/blob/master/lib/Cake/I18n/Multibyte.php#L59 ← oh! )

By the way, your suggestion code got syntax error?
"static $mbStrlen = function_exists("mb_strlen");" //Parse error: syntax error, unexpected '(',
It seems static var can't assign functions'return direct.

You are right - The only point I mean is that a static buffered check against "mb_strlen" could be enough to use "mb_strlen" or "strlen".
What is the performance difference of "mb_strlen($str, '8bit')" against "strlen($str)" ?